U.S. patent application number 15/527284 was filed with the patent office on 2017-12-21 for scoring of tumor infiltration by lymphocytes.
The applicant listed for this patent is THE INSTITUTE OF CANCER RESEARCH ROYAL CANCER HOSPITAL. Invention is credited to Yinyin Yuan.
Application Number | 20170365053 15/527284 |
Document ID | / |
Family ID | 52292441 |
Filed Date | 2017-12-21 |
United States Patent
Application |
20170365053 |
Kind Code |
A1 |
Yuan; Yinyin |
December 21, 2017 |
SCORING OF TUMOR INFILTRATION BY LYMPHOCYTES
Abstract
A method of providing a prognosis in a cancer patient comprising
analysing a tumour image to calculate a metric of immune
infiltration for the tumour, and a method of analysing a tumour
image.
Inventors: |
Yuan; Yinyin; (London,
GB) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
THE INSTITUTE OF CANCER RESEARCH ROYAL CANCER HOSPITAL |
London |
|
GB |
|
|
Family ID: |
52292441 |
Appl. No.: |
15/527284 |
Filed: |
November 24, 2015 |
PCT Filed: |
November 24, 2015 |
PCT NO: |
PCT/GB2015/053571 |
371 Date: |
May 16, 2017 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06T 2207/30096
20130101; G06T 2207/30068 20130101; G06T 7/0012 20130101; G01N
2800/52 20130101; G01N 33/57415 20130101 |
International
Class: |
G06T 7/00 20060101
G06T007/00; G01N 33/574 20060101 G01N033/574 |
Foreign Application Data
Date |
Code |
Application Number |
Nov 24, 2014 |
GB |
1420859.9 |
Claims
1. A method of measuring immune infiltration in a tumour, the
method comprising: providing an image of the tumour in which
lymphocytes and cancer cells have been identified; obtaining a
lymphocyte-to-cancer measurement for each lymphocyte; classifying a
subset of the lymphocytes as intra-tumour lymphocytes according to
their lymphocyte-to-cancer measurement; quantifying the
intra-tumour lymphocytes and the cancer cells in the tumour image;
calculating the intra-tumour lymphocyte ratio (ITLR) as the ratio
of intra-tumour lymphocytes to cancer cells, wherein the ITLR is a
measurement of immune infiltration in the tumour.
2. The method according to claim 1, wherein the step of obtaining a
lymphocyte-to-cancer measurement for each lymphocyte comprises:
applying a density estimate to obtain a model of the cancer cell
density; and determining the proximity of each lymphocyte to cancer
cell density.
3. The method according to claim 1, wherein a lymphocyte is
classified as an intra-tumour lymphocyte if its
lymphocyte-to-cancer measurement is above a threshold value.
4.-5. (canceled)
6. The method according to claim 1, further comprising a step of
identifying the lymphocytes and cancer cells in an image of the
tumour by automated image analysis to provide an image of the
tumour in which lymphocytes and cancer cells have been
identified.
7. The method according to claim 1 wherein the tumor is a tumor
sample from a cancer patient having breast cancer, ovarian cancer,
colorectal cancer, melanoma or non-small cell lung cancer.
8. The method of according to claim 7 wherein the cancer patient
has breast cancer which is triple negative breast cancer.
9.-11. (canceled)
12. A method of treating cancer in a cancer patient according to a
therapeutic regime, the method comprising analysing a tumour image
from the cancer patient according to the method of claim 1, and
treating the cancer patient according to the therapeutic regime
depending on whether the ITLR is below or above a predetermined
cut-off value.
13. The method according to claim 12, wherein the method further
comprises surgically resecting a tumour from the cancer patient and
measuring immune filtration in the surgically resected tumour.
14. The method of treating cancer according to claim 12, wherein
the cancer patient has triple negative breast cancer, wherein the
therapeutic regime comprises administration of a CTLA4 antagonist,
and wherein the cancer patient is treated according to the
therapeutic regime if the ITLR is above a predetermined cut-off
value.
15. The method according to claim 12, wherein the image of a tumour
is an image of a hematoxylin and eosin stained tumour section.
16.-18. (canceled)
Description
FIELD OF THE INVENTION
[0001] The present invention relates to tumour analysis, and to
cancer prognosis. In particular, the present invention relates to
methods of analysing tumours for determining a prognosis in
cancer.
BACKGROUND
[0002] Cancer is a complex and dynamic disease, and many different
ways of analysing and classifying tumours have been developed with
the aims of determining the degree of tumour progression or
invasiveness and the prognosis for the patient, and informing
treatment decisions.
[0003] Methods of analysing tumours include the assessment of cell
morphology in tumours (typically performed by pathologists),
measurement of gene expression in tumours (e.g. by microarray
analysis), determination of gene mutation status in tumour cells,
and evaluating protein expression within tumours (e.g. by
immunohistochemical assessment of tumour sections). These methods
of analysing tumours are important not only for predicting clinical
outcome, but also for informing decisions on patient therapy.
[0004] More recently, it has become apparent that the immunological
status of tumours can yield useful prognostic information.
Accumulating evidence supports the clinical significance of immune
response in many cancer types (Galon et al. 2006, Denkert et al.
2010, Loi et al.). Consistent studies have reported associations
between immune activity and disease outcome as well as treatment
response (Galon et al. 2006, Denkert et al. 2010, Loi et al., Liu
et al., Lee et al., DeNardo et al.).
[0005] Furthermore, increasing evidence from clinical trials
supports the potential of therapies that target immune activity in
certain types of cancer (Robert et al., Stagg et al.). This is
perhaps best exemplified in late stage melanoma where recent
clinical trials have shown an increased survival advantage in
patients receiving the monoclonal antibody ipilimumab, which
targets the CTLA4 protein receptor that is expressed on the surface
of T cells (Robert et al.). This has led to the development of more
standardised methods of characterising tumour immune infiltrate in
cancers such as the "immunescore" that aims to quantify the in situ
immune infiltrate in addition to standardised clinical parameters
to aid prognostication and patient selection for immunotherapy in
colorectal cancers (Galon et al. 2014).
[0006] However, to facilitate the standardisation and
reproducibility of scoring immune infiltration, objective
approaches are urgently needed (Galon et al. 2014). Furthermore,
such approaches need to account for the complexity of immune
infiltration into tumours. Abundance, spatial heterogeneity and
type of immune cells are the key parameters of immune infiltration
(Galon et al. 2014, Fridman et al.). For example, the spatial
locations of immune cells have been shown to be useful in
predicting the prognosis of colorectal cancer (Galon et al. 2006).
Indeed the pathological "immunescore" is based on the numeration of
two lymphocyte populations (CD8+ and CD45RO+ cells), both in the
core of the tumour and in the invasive margin that maximises the
prognostic power (Galon et al. 2014).
[0007] Similarly, large-scale studies of breast cancer have
demonstrated that pathological assessment of tumour-infiltration
lymphocytes based on Hematoxylin & Eosin (H&E) stained core
biopsies is a significant predictor for response to neoadjuvant
chemotherapy in 1,058 breast cancer samples (Denkert et al. 2010).
Recently, a prospective study demonstrated that in HER2-negative
breast cancer stromal lymphocytes can be an independent predictor
of response to neoadjuvant chemotherapy (Issa-Nummer et al.). Thus,
the spatial organisation of lymphocytic infiltration in the context
of nearby cancer cells is an important clinicopathological feature
of tumours.
[0008] In triple-negative breast cancer (TNBC) an active immune
response has been associated with favourable prognosis (Loi et al.,
Liu et al., Denkert et al.). A large-scale immunohistochemistry
study of 3,400 breast cancer samples has showed that TNBC is the
only subtype of breast cancer to demonstrate a significant link
between CDS-positive immune cells and a good prognosis (Liu et
al.). Assessment of lymphocytic infiltration based on whole-tumour
H&E sections has been associated with favourable outcome in 256
patients after anthracycline-based chemotherapy (Loi et al.). A
recent prospective study showed that the presence of
tumour-infiltrating lymphocytes in residual tumours after
neoadjuvant chemotherapy is predictive of good prognosis in TNBC
(Dieci et al.). Given the current lack of targeted molecular
treatment and poor clinical outcome of TNBC, this may suggest new
therapeutic opportunities for this aggressive tumour type (Stagg et
al.). For instance, accumulating data suggest that anthracyclines
mediate their action through activation of CD8+ T-cell responses,
hence combination with certain immunotherapies could be especially
effective for TNBC (Stagg et al.).
[0009] However, despite these advances in understanding of the
importance of immune infiltration in cancer, there is a lack of
reproducible approaches to objectively assess immune infiltration
based on pathological sections.
SUMMARY OF THE INVENTION
[0010] Lymphocytic infiltration in tumours is often associated with
a favourable prognosis and predicts response to chemotherapy in
many cancer types. However, it is not well understood because the
high levels of spatial and molecular heterogeneity within tumours
make it difficult to analyse by traditional pathological
assessment.
[0011] Identification of cell types by pathologists in the
assessment of immune infiltration provides qualitative information
on coarse ordinal scales. Such information is poorly suited to
analysing large data collections, partly because the high amount
human input required renders large scale studies time-consuming and
expensive, partly because the subjective nature of the assessment
causes an unacceptable degree of variability in the information,
and partly because the qualitative data generated do not lend
themselves to statistical analysis.
[0012] The inventor has devised a robust and reproducible method
for objectively assessing immune infiltration in tumours. The
method is performed on a tumour image in which lymphocytes and
cancer cells have been identified.
[0013] The method may be performed on images of hematoxylin &
eosin (H&E) stained tumour sections. H&E stained sections,
and images of H&E stained sections, are often readily available
as part of data sets collected for cancer study groups such as the
METABRIC group (Curtis, 2012) and the Cancer Genome Atlas (TCGA)
group (TCGA, 2012), which makes the methods of the present
invention readily adaptable for use in analysing tumours from a
variety of cancer types. The method may comprise a step of treating
a tumour section with a stain, such as H&E, wherein the
presence of subcellular structures such as nuclei creates complexes
between the stain and the subcellular structure.
[0014] An aspect of the present invention provides a method of
measuring immune infiltration in a tumour. In particular, there is
provided a method of determining an objective measurement of immune
infiltration in a tumour, referred to herein as the ITLR. The ITLR
(Intra-Tumour Lymphocyte Ratio) is the ratio of intra-tumour
lymphocytes to cancer cells in the tumour expressed as a decimal
fraction. For example a ratio of 11 intra-tumour lymphocytes to
1000 cancer cells corresponds to an ITLR of 0.011.
[0015] Accordingly, an aspect of the invention provides a method of
measuring immune infiltration in a tumour, the method comprising:
[0016] providing an image of the tumour in which lymphocytes and
cancer cells have been identified; [0017] obtaining a
lymphocyte-to-cancer measurement for each lymphocyte;
[0018] classifying a subset of the lymphocytes as intra-tumour
lymphocytes according to their lymphocyte-to-cancer ratio; [0019]
quantifying the intra-tumour lymphocytes and the cancer cells in
the tumour image;
[0020] calculating the intra-tumour lymphocyte ratio (ITLR) as the
ratio of intra-tumour lymphocytes to cancer cells, wherein the ITLR
is a measurement of immune infiltration in the tumour.
[0021] A further aspect of the present invention provides a method
of determining a cut-off value for ITLR for use in determining a
prognosis in cancer, wherein an ITLR below the cut-off value
indicates a poor prognosis. The method comprises determining the
ITLR for a plurality of tumours, wherein each tumour is from a
respective cancer patient in a cohort of cancer patients, and
selecting a cut-off value for the ITLR wherein patients with an
ITLR lower than the cut-off value have a worse prognosis compared
with patients with an ITLR equal to or higher than the cut-off
value.
[0022] Accordingly, an aspect of the invention provides a method of
determining an ITLR cut-off value for a cancer type or subtype, for
use in providing a prognosis in a cancer patient having that cancer
type, the method comprising: [0023] measuring immune infiltration
in a tumour from each member of a cohort of cancer patients having
the cancer type or subtype according to the methods described
herein, thereby calculating the ITLR for each tumour; [0024]
relating the ITLR for each tumour to the clinical outcome of each
cancer patient in the cohort of cancer patients; and [0025]
selecting a cut-off value for ITLR, wherein an ITLR equal to or
below the cut-off value is associated with a significantly worse
clinical outcome in the cohort of cancer patients than an ITLR
above the cut-off value
[0026] A further aspect of the present invention provides a method
of providing a prognosis in cancer. In particular, there is
provided a method of using ITLR as a prognostic biomarker for a
cancer patient. The method may comprise measuring the ITLR of a
tumour from a cancer patient and using the ITLR to determine a
prognosis for the patient. The method may comprise determining the
ITLR in a tumour from a cancer patient and using the ITLR to
determine a prognosis for the patient, wherein an ITLR below a
predetermined cut-off value indicates a poor prognosis.
[0027] Accordingly, an aspect of the invention provides a method of
providing a prognosis in a cancer patient, the method comprising:
[0028] measuring immune infiltration in a tumour from the cancer
patient according to the methods described herein, thereby
calculating the ITLR for the tumour, [0029] wherein an ITLR below a
predetermined ITLR cut-off value indicates a poor prognosis.
[0030] The present invention further provides a method of treating
cancer in a patient, the method comprising determining the ITLR in
a tumour from the patient or requesting a test providing the
results of an analysis to determine the ITLR in a tumour from the
patient, and treating the patient according to a therapeutic regime
depending on whether the ITLR is equal to or below, or above, a
predetermined-cut-off value.
SUMMARY OF THE FIGURES
[0031] FIG. 1. Intra-tumour heterogeneity of cancer cell and
lymphocyte distributions.
[0032] A. 3D landscapes illustrating the spatial heterogeneity of
cancer cells and lymphocytes in an H&E breast whole-tumour
section. The height of the hills in the 3D landscape represents the
density of cells. B. Combined analysis of the spatial distribution
of cancer and lymphocytes can lead to quantification of lymphocytic
infiltration. Shown are a small H&E image and the corresponding
3D cancer density map, which facilitate the measurement of spatial
proximity to cancer for every single lymphocyte in the image.
[0033] FIG. 2. Quantifying the intra-tumour heterogeneity of
lymphocytic infiltration.
[0034] A. Schematic depiction of the computational pipeline
exemplified with a small region of a breast cancer H&E section:
H&E image; classified cells using automated image analysis; a
map of cancer density based on image analysis result to quantify
cancer-immune spatial relationships. B. Discovery of three
categories of lymphocytes with unsupervised clustering based on the
spatial proximities of lymphocytes to cancer in a subset of TNBC
samples. These data were then used to predict the categories of all
lymphocytes in all TNBC samples. C. Optimal number of cluster K as
suggested by BIC over 200 random sampling are 3 in 97% of the time
and 5 (3%). BIC curves for the 200 sampling are showed on the left,
and boxplot showing means of clusters for K=3 solutions in 200
sampling on the right. D. Illustration of the distance to the
nearest cancer cell d.sub.min and the distance to the centroid of
convex hull region formed by 10 nearby cancer cells d.sub.centroid
E. Boxplots to show the differences among lymphocyte classes in
terms of d.sub.min and d.sub.centroid (p-values by t-test). F.
Scatter plot showing drain and d.sub.centroid for 1,000 randomly
selected lymphocytes, coloured based on the three classes; dashed
ellipses showing three clusters fitted to d.sub.min and
d.sub.centroid.
[0035] FIG. 3. A representative example illustrating three classes
of lymphocytes in cancer density map of a tumour (middle
section).
[0036] A. Density map of cancer and the spatial distribution of
three classes of lymphocytes (spatial points coloured according to
the classes). Black contour lines denote cut-off thresholds for the
three classes of lymphocytes according to cancer density. B.
Histogram showing the three types of lymphocytes in this sample. C.
A higher resolution image of a region in this sample; colour codes
follow A.
[0037] FIG. 4. Association between ITLR and clinical parameters of
TNBC.
[0038] A. Proportions of three classes of lymphocytes in 181 TNBCs.
B. Triangle plot to show the lymphocyte composition for each tumour
(each black dot represents a tumour; thin lines mark the 50% of
corresponding axis). C. Boxplot to show correlation between
pathological scores and ITLR; p-value from JT-test; n=patient
number is each group; whiskers extend to 1.5 interquantile range.
D. Association between ITLR and tumour size, node status and TP53
mutations; whiskers extend to 1.5 interquantile range. E.
Distribution of ITLR in two cohorts with optimal cut-offs marked as
dashed red lines. F. Kaplan-Meier curves to illustrate the
disease-specific survival probabilities of patient groups in two
TNBC cohorts stratified by ITLR using the cut-off selected in
Cohort 1. Numbers in the legend show the number of patients in each
group and numbers in the bracket show the number of
disease-specific deaths. G. Using Cohort 2 as the discovery cohort
and Cohort 1 as the validation cohort yielded similar optimal
cut-off.
[0039] FIG. 5. Comparison of ITLR with other immune signatures.
[0040] The optimal cut-off were selected in Cohort 1 and tested in
Cohort 2 for A. image-based lymphocyte abundance (Lym); B. gene
expression immune signature by Calabro et al. (18); C. Ascierto et
al. (19); D. IL8 signature (20); E. CXCL13 expression. F. Comparing
optimal cut-offs selected in two cohorts. Data were centred at 0
and scaled to have standard deviation 1 and cut-offs were mapped to
the centred, scaled data. Signatures close to the diagonal line
have similar cut-offs in two cohorts.
[0041] FIG. 6. ITLR-associated gene modules.
[0042] A. Kaplan-Meier curves to illustrate differences in
disease-specific survival of patient groups of equal sizes
stratified based on the expression of key genes in three modules.
B. Kaplan-Meier curves to illustrate differences in
disease-specific survival of patient groups stratified with CTLA4
expression by the lower 25, middle 50 and higher 25 percentiles,
ITLR, and CTLA4 and ITLR combined. Survival difference between CTLA
low and high stratification within the ITLR high group is given as
a p-value.
[0043] FIG. 7. Kaplan-Meier curves to illustrate the
disease-specific survival probabilities of patient groups in in two
TNBC cohorts stratified by ATLR (Adjacent) and DTLR (Distal).
[0044] The signatures were dichotomised using a cut-off selected
over a range of percentiles based on Cohort 1 (the left and middle
columns) and tested in Cohort 2 (the right column). Dashed lines in
the plots on the left marks the significance threshold of p=0.05,
and solid vertical lines show the best cut-offs. For the
Kaplan-Meier curves, the numbers in the legend show the number of
patients in each group and numbers in the bracket show the number
of disease-specific deaths.
[0045] FIG. 8. Kaplan-Meier curves to illustrate the
disease-specific survival probabilities of patient groups in in two
TNBC cohorts stratified by ATLR (Adjacent) and DTLR (Distal).
[0046] The signatures were dichotomised using a cut-off selected
over a range of percentiles based on Cohort 2 (the left and right
columns) and tested in Cohort 1 (the middle column).
[0047] FIG. 9. Kaplan-Meier curves to illustrate the
disease-specific survival probabilities of patient groups in in two
TNBC cohorts stratified by nine immune signatures.
[0048] The signatures were dichotomised using a cut-off selected
over a range of percentiles based on Cohort 1 (the left and middle
columns) and tested in Cohort 2 (the right column). Dashed lines in
the plots on the left marks the significance threshold of p=0.05,
and solid vertical lines show the best cut-offs. For the
Kaplan-Meier curves, the numbers in the legend show the number of
patients in each group and numbers in the bracket show the number
of disease-specific deaths.
[0049] FIG. 10. Kaplan-Meier curves to illustrate the
disease-specific survival probabilities of patient groups in in two
TNBC cohorts stratified by nine immune signatures.
[0050] The signatures were dichotomised using a cut-off selected
over a range of percentiles based on Cohort 2 (the left and right
columns) and tested in Cohort 1 (the middle column). Dashed lines
in the plots on the left marks the significance threshold of
p=0.05, and solid vertical lines show the best cut-offs. For the
Kaplan-Meier curves, the numbers in the legend show the number of
patients in each group and numbers in the bracket show the number
of disease-specific deaths.
[0051] FIG. 11. Scatter plots to show correlation between ITLR and
expression of ITLR-associated genes in TNBC.
[0052] FIG. 12. Compare the prognostic value of top 100
ITLR-associated genes and ITLR by including both in multivariate
Cox analysis model, one gene at a time.
[0053] Each point denotes analysis for one gene, plotted values are
log(log rank p-value) for the analysis.
[0054] FIG. 13. Kaplan-Meier curves illustrating differences in
disease-specific survival of TNBC patients stratified with other
known parameters including PAM50 (Perou et al., 2000), pathological
assessment of lymphocytic infiltration (LI), tumour size, and
grade.
[0055] FIG. 14. Kaplan-Meier curves illustrating differences in
5-year overall survival of ovarian cancer patients of patient
groups stratified by ITLR, by "Lym" (Yuan et al., 2012), by lymPath
(lymphocyte abundance assessed by pathologist), tumour grade,
histologic type, or tumour staging.
[0056] FIG. 15. A. Histogram showing clustering of cells from TNBC
tumours into cluster having relatively high CTLA4 expression and
cluster having relatively low CTLA4 expression. B. Kaplan-Meier
curve illustrating difference in survival between patients having
low CLTA4 expression and high CTLA4 expression.
DETAILED DESCRIPTION OF THE INVENTION
[0057] Certain aspects and embodiments of the invention will now be
illustrated by way of example and with reference to the figures
described above.
[0058] The inventor has devised a novel way of statistically
modelling the spatial heterogeneity of lymphocytes in tumours,
which enables determination of a quantitative measurement of immune
infiltration (ITLR). ITLR (intra-tumour lymphocyte ratio) is the
ratio of intra-tumour lymphocytes to cancer cells in a tumour. This
quantitative measurement of immune infiltration (ITLR) has improved
predictive power in cancer prognosis compared with previous
indicators of immune infiltration. This measurement of immune
infiltration was developed based on a study of tumours from triple
negative breast cancer (TNBC) patients from the METABRIC dataset,
but is more generally useful in and applicable to other breast
cancer sub-types and other cancer types. The generalizable nature
of ITLR is demonstrated by the data herein showing that ITLR is
also a prognostic indicator in ovarian cancer.
[0059] Described herein is the first study to statistically
identify categories of lymphocytes based on tumour spatial
heterogeneity and demonstrate their clinical implications using
samples from a large number of patients. This enables a way of
modelling spatial heterogeneity in tumours which addresses the need
for measuring heterogeneity of lymphocytic infiltration in tumours.
The ability to generate reproducible, quantitative scores provides
new opportunities for incorporating immune infiltration into
staging of cancer (i.e. grading of tumours), as in the use of
immunoscore for colorectal cancer (Galon, 2014).
[0060] The present invention provides a method of measuring immune
infiltration in tumours. In particular, there is provided a method
of determining an objective measurement of immune infiltration in a
tumour (ITLR), which measurement is the ratio of intra-tumour
lymphocytes to cancer cells.
[0061] Accordingly, an aspect of the present invention provides a
method of measuring immune infiltration in a tumour, the method
comprising: [0062] providing an image of the tumour in which
lymphocytes and cancer cells have been identified; [0063] obtaining
a lymphocyte-to-cancer measurement for each lymphocyte;
[0064] classifying a subset of the lymphocytes as intra-tumour
lymphocytes according to their lymphocyte-to-cancer ratio; [0065]
quantifying the intra-tumour lymphocytes and the cancer cells in
the tumour image;
[0066] calculating the intra-tumour lymphocyte ratio (ITLR) as the
ratio of intra-tumour lymphocytes to cancer cells, wherein the ITLR
is a measurement of immune infiltration in the tumour.
[0067] The methods described herein may be performed using a tumour
image in which lymphocytes and cancer cells have been identified.
The lymphocytes and cancer cells may have been identified by
automated image analysis.
[0068] The methods described herein may further comprise a step of
identifying cancer cells and lymphocytes in a tumour image by
automated image analysis. The methods may comprise steps of
generating a tumour image and then identifying lymphocytes and
cancer cells in the tumour image by automated image analysis.
[0069] The step of identifying cancer cells and lymphocytes in a
tumour by automated image analysis may be based on the different
nuclear morphologies of cancer cells and lymphocytes. This step may
be performed on tumour sections, such as whole-tumour section
slides. The tumour section may be H&E stained. The types and/or
spatial locations of at least about 10,000 cells may be recorded in
this step. The types and/or spatial locations of at least about
20,000, at least about 50,000, at least about 90,000, at least
100,000, at least about 110,000, about 10,000 to 150,000, about
50,000 to 120,000, or about 100,000 to 120,000 cells may be
recorded in this step. The types and/or spatial locations of, or
about 90,000, about 100,000, or about 110,000 cells may be recorded
in this step. The cells may be lymphocytes. This step may use any
automated image analysis tool capable of identifying lymphocytes
and cancer cells. The automated image analysis tool may be the tool
disclosed in Yuan et al, 2012.
[0070] The image analysis tool disclosed in Yuan et al, 2012, which
is hereby incorporated by reference in its entirety, identifies
cancer, lymphocytes and stromal cells encompassing fibroblasts and
endothelial cells based on their nuclear morphologies in H&E
whole-tumour section slides. The main component of this tool is a
classifier trained by pathologists over randomly selected tumour
regions and validated in 564 breast tumours with 90% accuracy. The
image analysis tool described in Yuan et al classifies cells into
three categories: cancer, lymphocyte or stromal based on
morphological features using a support vector machine.
[0071] The image analysis tool described in Yuan et al, 2012
identified cancer cells by their typically large (>10 .mu.m),
round nuclei. The stromal class was trained on spindle-shaped
stromal cell nuclei (likely to be fibroblasts) and may encompass
other stromal cells with similar morphology, such as endothelial
cells. The lymphocyte class was trained on immune cells with the
distinctive morphology of lymphocytes: small (<8 .mu.m), dark
nuclei and not much cytoplasm.
[0072] The image analysis tool described in Yuan et al was trained
using breast tumour images. Automated image analysis tools, such as
those described in Yuan et al 2012, can be trained in cancer types
other than breast cancer (including those cancer types and subtypes
mentioned herein) in order to identify lymphocytes and cancer cells
in the tumours of other cancer types.
[0073] Various automated image analysis tools are known in the art.
For example the tools described in Failmezger et al (CRImage)
particular Janowczyj et al, and Basavanhally et al, which are
hereby incorporated by reference in their entirety. Any such tool
may be suitable for, or adapted for, use in the methods described
herein.
[0074] As a result of automated image analysis, the types and
spatial locations of a large number of cells are recorded in every
tumour image. The automated image analysis may enable the mapping
of spatial distributions of all, or essentially all, cancer cells
and lymphocytes within a tumour image.
[0075] Following a step of identifying the cancer cells and
lymphocytes using automated image analysis, the spatial
relationships of lymphocytes and cancer cells are analysed.
[0076] The methods described herein comprise a step of obtaining a
lymphocyte-to-cancer measurement for each lymphocyte. This provides
a quantitative measurement of each lymphocyte's proximity to cancer
cells and spatial location relative to cancer cells.
[0077] The step of obtaining a lymphocyte-to-cancer measurement for
each lymphocyte may be carried out using the statistical pipeline
exemplified in FIG. 1B. First, to globally profile the spatial
distribution of the cancer cells, the cancer cell density is
quantified, for example using a kernel estimate (Hastie et al,
2001). Alternatively, a mean shift estimate (Cheng, 1995) or scale
space (Witkin, 1983) estimate may be used. This builds a `cancer
landscape` where hills indicate tumour regions densely populated
with cancer cells. The height of a hill thus correlates with cancer
density (tumour density) at a specific location in the tumour (FIG.
1B). Secondly, for every lymphocyte, its spatial proximity to
cancer is directly quantified with the cancer density landscape at
its specific location to give a "lymophocyte-to-cancer" measurement
for each lymphocyte. Thus a quantitative measurement of the spatial
proximity to cancer cells is obtained for each lymphocyte (FIG.
1B).
[0078] In the studies described herein (see Experimental), cancer
cells and lymphocytes were identified, and then their spatial
relationships were quantified using a kernel density method. Then,
using unsupervised learning, three categories of lymphocytes
(intra-tumour, adjacent-tumour and distal tumour) were identified
based on their spatial proximities and spatial positioning relative
to cancer cells. These lymphocyte categories are consistent with a
pathological quantification scheme that considers intratumoral,
adjacent stroma and distant stroma compartments (Mahmoud, 2011).
Statistically, these clusters are stable, reported as the optimal
clustering solution 97% of the time upon repeated sampling.
[0079] Accordingly, the methods described herein may comprise a
step of obtaining a lymphocyte-to-cancer measurement for each
lymphocyte by using a density estimate, such as a kernel estimate,
to model the spatial distribution of the cancer cells. The method
then comprises a step of determining the proximity of each
lymphocyte to cancer by determining the cancer cell density at the
location of each lymphocyte, to give a lymphocyte-to-cancer
measurement for each lymphocyte. The lymphocytes are then clustered
according to their lymphocyte-to-cancer measurements. An
unsupervised learning method, such as Gaussian mixture clustering,
may be used to cluster lymphocytes according to their proximity to
cancer. The number of clusters may be 2, 3, 4 or more.
[0080] In the TNBC study described herein (see Experimental), when
lymphocytes were clustered according to their lymphocyte-to-cancer
measurements the number of clusters was three (k=3), corresponding
to intra-tumour lymphocytes (ITL), adjacent tumour lymphocytes
(ATL) and distal tumour lymphocytes (DTL).
[0081] In TNBC, lymphocytes having a lymphocyte-to-cancer
measurement above the threshold value of 0.10507473 were classified
as ITLs, lymphocytes having a lymphocyte-to-cancer measurement
below the threshold value of 0.10507473 and above the threshold
value of 0.03662728 were classified as ATLs, and lymphocytes having
a lymphocyte-to-cancer measurement below the threshold value of
0.03662728 were classified as DTLs. In determining the ITLR, the
important distinction is between intra-tumour lymphocytes (ITLs)
and non intra-tumour lymphocytes (non-ITLs). Thus in TNBC,
lymphocytes having a lymphocyte-to-cancer measurement equal to or
above the threshold value of 0.10507473 were classified as ITLs,
and the remaining lymphocytes were classified as non-ITLs.
[0082] In the ovarian cancer study described herein, when
lymphocytes were clustered according to their lymphocyte-to-cancer
measurements the number of clusters was two (k=2), corresponding to
intra-tumour lymphocytes and non-intra-tumour lymphocytes.
[0083] In ovarian cancer, lymphocytes having a lymphocyte-to-cancer
measurement above the threshold value of 0.03114299 were classified
as ITLs. Lymphocytes having a lymphocyte-to-cancer measurement
below this threshold value were classified as non-ITLs.
[0084] As an alternative to using cancer density at a lymphocyte
location to give a lymphocyte-to-tumour measurement that is
indicative of lymphocyte proximity to cancer (i.e. lymphocyte
closeness to cancer), the step of obtaining a lymphocyte-to-cancer
measurement for each lymphocyte may be carried out based on a
distance measure between a lymphocyte and one or more cancer cells,
such as the Euclidean distance. The lymphocytes are then clustered
according to their lymphocyte-to-cancer measurements, as described
above, for example using an unsupervised learning method, such as
Gaussian mixture clustering. In this context, where the
lymphocyte-to-cancer measurement is indicative of distance from
(rather than proximity to) cancer, a lymphocyte may be classified
as an ITL if it has lymphocyte-to-cancer measurement below a
threshold value.
[0085] The methods described herein may comprise classifying
lymphocytes as intra-tumour lymphocytes. That is, the methods may
comprise classifying a subset of cells identified as lymphocytes in
the tumour image as intra-tumour lymphocytes. Classifying
lymphocytes may comprise determining whether the
lymphocyte-to-cancer measurement is above a certain threshold
value. The threshold value, for example in TNBC, may be around 0.1,
around 0.105 or around 0.10507473. The threshold value, for example
in ovarian cancer, may be around 0.03, around 0.0311, or around
0.03114299.
[0086] The methods described herein may comprise determining a
threshold value for a lymphocyte-to-cancer measurement, for use in
classifying a lymphocyte as an intra-tumour lymphocyte or a
non-intra-tumour lymphocyte. For example, where the
lymphocyte-to-cancer measurement is indicative of lymphocyte
proximity to cancer, the lymphocyte may be classed as an
intra-tumour lymphocyte if it has a lymphocyte-to-cancer
measurement above the lymphocyte-to-cancer measurement threshold
value. Determining a threshold value for a lymphocyte-to-cancer
measurement may comprise determining lymphocyte-to-cancer
measurements for a population of lymphocytes and clustering the
lymphocytes by unsupervised learning, and taking the minimum value
of the most cancer proximal cluster (the cluster with the highest
measurements) as the threshold value for classifying intra-tumour
lymphocytes. A lymphocyte may be classified as an intra-tumour
lymphocyte if it has a lymphocyte-to-cancer measurement above (or
equal to or above) the threshold value.
[0087] Determining the threshold value may further comprise testing
the stability of the clustering by sampling the population of
lymphocytes, clustering the sampled population of lymphocytes and
determining that the cluster solution (k=x where x is the number of
clusters) is stable. The number of clusters is stable where k for
the sampled population is the same for 200 repeated samples at
least 90%, at least 95% or at least 97% of the time.
[0088] Furthermore, the inventor has shown significant differences
between lymphocyte categories both in spatial distance to the
nearest cancer cell and spatial positioning of surrounding cancer
cells, supporting their biological relevance. For instance, in the
presently disclosed study of tumours from TNBC patients from the
METABRIC dataset, an intra-tumour lymphocyte is on average 7 .mu.m
away from a cancer cell and 3 .mu.m from the centroid of convex
hull region formed by nearby cancer cells. An adjacent-tumour
lymphocyte may be also close to the nearest cancer cells but would
be further away from the centroid of convex hull region because it
is not surrounded by cancer cells. Thus, the new classification
approach disclosed herein is based on spatial measures that account
for spatial positioning of cancer cells whilst being
computationally efficient enough to analyse whole-tumour sections.
Compared to a previously reported measure of lymphocyte abundance
as a direct output from image analysis (Yuan, 2012), an advantage
of this new approach is that it accounts for the spatial
heterogeneity of immune infiltration, which is recognised as an
important property of immune infiltration (Galon, 2006) but rarely
quantitatively analysed.
[0089] Following the step of classifying lymphocytes as
intra-tumour lymphocytes, the ratio of intra-tumour lymphocytes to
cancer cells is calculated. This ratio is the ITLR (the
intra-tumour lymphocyte ratio), which is an objective and
quantitative measurement of immune infiltration in tumours. The
ITLR is the ratio of intra-tumour lymphocytes to cancer cells in
the tumour expressed as a decimal fraction. For example, an ITLR of
0.011 represents a 1.1% of intra-tumour lymphocytes to cancer cells
i.e. a ratio of 11 intra-tumour lymphocytes to 1000 cancer
cells.
[0090] The inventor has shown that ITLR is a robust and powerful
prognostic indicator in triple negative breast cancer (TNBC), as
discussed below, and also in ovarian cancer. Since immune
infiltration is implicated in many cancer types, as discussed in
more detail below, including breast cancer, ovarian cancer,
colorectal cancer (Galon, 2014), melanoma and non-small cell lung
cancer, ITLR may also be used as a prognostic indicator in various
cancer types.
[0091] For prognosis in TNBC, the ITLR cut-off of 0.011 was
selected based on tumour images from the METABRIC cohort. Patients
whose tumours had an ITLR below the cut-off value of 0.011 had a
significantly worse clinical outcome in terms of disease-specific
survival compared with patients whose tumours had an ITLR above the
cut-off value.
[0092] For prognosis in ovarian cancer, the ITLR cut-off of 0.06086
was selected based on tumour images from an unpublished tumour
cohort. Patients whose tumours had an ITLR below the cut-off value
had a significantly worse clinical outcome in terms of overall
survival compared with patients whose tumours had an ITLR above the
cut-off value.
[0093] An aspect of the present invention provides a method of
determining a cut-off value for ITLR for use in determining a
prognosis in cancer, wherein an ITLR below the cut-off value
indicates a poor prognosis. The method comprises determining the
ITLR for a plurality of tumours, wherein each tumour is from a
respective cancer patient in a cohort of cancer patients, and
selecting a cut-off value for the ITLR wherein patients with an
ITLR equal to or below the cut-off value have a significantly worse
prognosis compared with patients with an ITLR above the cut-off
value.
[0094] Accordingly, an aspect of the present invention provides a
method of determining an ITLR cut-off value for a cancer type or
subtype, for use in providing a prognosis in a cancer patient
having that cancer type or subtype, the method comprising: [0095]
measuring immune infiltration in a tumour from each member of a
cohort of cancer patients having the cancer type or subtype
according to the method of claim 1, thereby calculating the ITLR
for each tumour; [0096] relating the ITLR for each tumour to the
clinical outcome of each cancer patient in the cohort of cancer
patients; and [0097] selecting a cut-off value for ITLR, wherein an
ITLR equal to or below the cut-off value is associated with a
significantly different clinical outcome in the cohort of cancer
patients than an ITLR above the cut-off value.
[0098] An ITLR equal to or below the cut-off value may be
associated with a significantly worse clinical outcome than an ITLR
above the cut-off value. An ITLR equal to or below the cut-off
value may be associated with a significantly better clinical
outcome than an ITLR above the cut-off value.
[0099] The selection of the cut-off value for ITLR serves to
dichotomise the continuous range of ITLR values for the tumour
images from a patient cohort. The ITLR cut-off value is selected
such that there is a significant difference in clinical outcome
between patients with an ITLR below the cut-off and patients with
an ITLR above the cut-off value. In general, the ITLR cut-off value
is selected such that patients having an ITLR below or equal to the
cut-off value (i.e. patients having a tumour with an ITLR below or
equal to the cut-off value) have a significantly worse prognosis
than patients having an ITLR that is above the cut-off value (i.e.
patients having a tumour with an ITLR equal to or above the cut-off
value).
[0100] In the context of the present invention a significant
difference in prognosis refers to a clinical outcome that is
significantly different according to the Log rank test. Preferably
p<0.0500, p<0.0250, p<0.0100, p<0.0090, p<0.0065,
p<0.0010, or p<0.0001 according to the log rank test.
[0101] Selection of the ITLR cut-off value may comprise identifying
an ITLR value wherein about 20% to 80% of the patient cohort has an
ITLR below that value. Selection of the ITLR cut-off value may
comprise identifying an ITLR value wherein about 20% to 80% of the
patient cohort has an ITLR below that value and wherein patients
having an ITLR below the cut-off value have a significantly worse
prognosis than patients having an ITLR that is above the cut-off
value.
[0102] The clinical outcome may be disease-specific survival,
disease free survival, overall survival, relapse-free survival,
progression-free survival, survival rate or survival time. The
clinical outcome may be disease-specific survival. Disease-specific
survival may be defined with time as maximum 5 years or 10 years
from diagnosis and event as death due to cancer (the 5 year disease
specific survival and 10 year disease specific survival
respectively). Overall survival may be defined with time as maximum
5 years or 10 years from diagnosis and event as death due to any
cause. Relapse-free survival may be defined with time as maximum 10
years from diagnosis and event as tumour relapse. A poor prognosis
refers to a prediction of a poor clinical outcome, whereas a
positive prognosis refers to a prediction of a positive clinical
outcome.
[0103] The methods described herein may use a cohort of cancer
patients from The Cancer Genome Atlas (TOGA) as the "discovery"
cohort. This dataset, with its H&E and matched molecular
profiling data will be an extremely useful cohort to validate the
utility of ITLR and to select and refine ITLR cut-off values for
use in prognostic and/or therapeutic methods. TCGA has chosen
cancers for study based on criteria that include poor prognosis and
overall public health impact and the availability of human tumour
and matched-normal tissue samples that meet TCGA standards for
patient consent quality and quantity.
[0104] The experiments disclosed herein show the utility of ITLR in
TNBC and in ovarian cancer. ITLR is a generalizable measure for
ITLs and will therefore be useful as a measure of immune
infiltration in other cancer types I subtypes, especially given
that manual assessment of ITLS has reported value in many cancer
types I subtypes.
[0105] As already mentioned above, immune infiltration is
implicated in many cancer types, including breast cancer, ovarian
cancer, colorectal cancer, melanoma and non-small cell lung cancer,
ITLR may also be used as a prognostic indicator in various cancer
types.
[0106] Immune infiltration is implicated in many cancers including
breast cancer (including breast ductal carcinoma breast and breast
lobular carcinoma) (Dieci 2014; Loi S 2013; Kruger J M 2013; Liu S,
2012; Ascierto M L 2012, Rody A, 2011; Mahmoud S M A, 2011; Denkert
C, 2010; Ueno T, 2000) central nervous system cancer (including
glioblastoma multiforme and lower grade glioma) (Kmiecik J, 2013;
Yang I, 2010; McNamara M G, 2014; Crane C A, 2014; Bambury R M;
Alexiou G A, 2013; Vauleon E, 2013) endocrine cancer (including
adrenocortical carcinoma, papillary thyroid carcinoma,
paraganglioma & pheochromocytoma) (Papewalis C; Huang C T;
Mukherji B) gastrointestinal cancer (including Cholangiocarcinoma,
Colorectal Adenocarcinoma, Liver Hepatocellular Carcinoma,
Pancreatic Ductal Adenocarcinoma, Stomach-Esophageal Cancer) (Kono
K, 20116; Wu G; Gao Q; Hiraoka N) gynecologic cancer (including
Cervical Cancer (Zhang Y, 2014; Ancuta E, 2009), Ovarian Serous
Cystadenocarcinoma (Townsend K N, 2013; Milne K 2009; Clarke B,
2009), Uterine Carcinosarcoma, Uterine Corpus Endometrial
Carcinoma) (Ohno S, 2004) head and neck cancer (including Head and
Neck Squamous Cell Carcinoma, Uveal Melanoma) (Spanos W C, 2009;
Pretscher D, 2009) hematologic cancer (including Acute Myeloid
Leukemia, thymoma, lymphoma) (Yong A S, 2011; Dave S S, 2004;) skin
cancer (including Cutaneous Melanoma) (Tjin E P, 2014; Erdag G,
2012; Bystryn J C, 1992; Halliday G, 1995) soft tissue cancer
(including Sarcoma) (Kim J R, 2016; Sorbye S W 2011; Fiorelli V,
1998) thoracic cancer (including Lung Adenocarcinoma, Lung Squamous
Cell Carcinoma, Mesothelioma) (Suzuki K, 2013; Welsh T J, 2005;
Villegas F R, 2002; Hegmans J P, 2006; Dieu-Nosjean M C) urologic
cancer (including Chromophobe Renal Cell Carcinoma, Clear Cell
Kidney Carcinoma, Papillary Kidney Carcinoma, Prostate
Adenocarcinoma, Testicular Germ Cell Cancer, Urothelial Bladder
Carcinoma) (Davidsson S, 2013; Thompson R H, 2007; Webster W S,
2006; Gannon P O, 2009; Sjodahl G, 2014).
[0107] The methods of the present invention may be applied in any
of the cancer types or subtypes mentioned above.
[0108] ITLR is an objective quantitative indicator of lymphocytic
infiltration in tumours. The inventor has shown the importance of
using such a quantitative measurement of lymphocytic infiltration
in predicting clinical outcome in cancer.
[0109] ITLR is a new spatial and quantitative measure of
intra-tumour lymphocytes (ITLs). This measure is a consistent,
stable and independent predictor of disease-specific survival
across two independent cohorts of 181 TNBC patients in total. This
measurement may use a cut-off of 0.011 (1.1% of intra-tumour
lymphocytes to cancer cells) that dichotomises the ITLR score. The
20% of TNBC patients with ITLR scores lower than this cut-off have
significantly worse disease-specific survival than patients with
higher scores, and this association is independent of standard
clinical parameters. Taken together, these data support the utility
of ITLR as a prognostic biomarker for cancer, including TNBC.
Accordingly, disclosed herein is an objective and fully automated
scoring system for the standardised assessment of immune
infiltration that can be used in the context of clinical trials and
subsequently aid the treatment decision making process.
[0110] Accordingly, an aspect of the present invention may further
comprise using ITLR as a prognostic biomarker. The method may
comprise measuring the ITLR of a tumour from a cancer patient and
using the ITLR to determine a prognosis for the patient. The method
may comprise determining the ITLR in a tumour from a cancer patient
and using the ITLR to determine a prognosis for the patient,
wherein an ITLR below a predetermined cut-off value indicates a
poor prognosis. The method may comprise determining the ITLR in a
tumour from a cancer patient and using the ITLR to determine a
prognosis for the patient, wherein an ITLR above a predetermined
cut-off value indicates a poor prognosis.
[0111] In particular, an aspect of the present invention provides a
method of providing a prognosis in a cancer patient, the method
comprising: [0112] measuring immune infiltration in a tumour from
the cancer patient according to a method described herein, thereby
calculating the ITLR for the tumour, [0113] wherein an ITLR below a
predetermined ITLR cut-off value indicates a poor prognosis.
[0114] An aspect of the present invention provides a method of
providing a prognosis in a cancer patient, the method comprising:
[0115] providing an image of the tumour in which lymphocytes and
cancer cells have been identified; [0116] obtaining a
lymphocyte-to-cancer measurement for each lymphocyte; [0117]
classifying a subset of the lymphocytes as intra-tumour lymphocytes
according to their lymphocyte-to-cancer ratio; [0118] quantifying
the intra-tumour lymphocytes and the cancer cells in the tumour
image; [0119] calculating the intra-tumour lymphocyte ratio (ITLR)
as the ratio of intra-tumour lymphocytes to cancer cells, wherein
the ITLR is a measurement of immune infiltration in the cancer
patient's tumour [0120] and wherein an ITLR below a predetermined
ITLR cut-off value indicates a poor prognosis.
[0121] The inventor has shown that ITLR is an independent predictor
of clinical outcome in cancer. That is, the ITLR is predictive of
clinical outcome without using any other biomarker (such as a gene
expression biomarker) or clinical indicator (such as tumour size).
For example, the inventor has shown that ITLR is an independent
predictor of clinical outcome in triple negative breast cancer
(TNBC) and in ovarian cancer. For example in the studies described
herein there was no correlation between ITLR and tumour size, node
status and TP53 mutation status (FIG. 4D), and so ITLR is
independent of such clinical indicators and biomarkers. Preferably,
if the ITLR is below a predetermined cut-off value (or equal to or
below a predetermined cut-off value), this indicates a poor
prognosis (i.e. a poor clinical outcome). A poor prognosis, or poor
clinical outcome, may be poor disease-specific survival.
[0122] In the present context, an ITLR below a predetermined
cut-off value may be referred to as a low ITLR, or ITLR-low.
Conversely an ITLR above a predetermined cut-off value may be
referred to as a high ITLR, or ILTR-high. A predetermined cut-off
value for an ITLR may simply be referred to herein as an ITLR
cut-off value.
[0123] A poor prognosis means that the patient has a worse
prognosis than a patient having an ITLR value above the cut-off
ITLR value. For example a poor prognosis may mean that the patient
is expected to have shorter disease-specific survival time than a
patient having an ITLR value above the cut-off ITLR value. A poor
prognosis may mean that the patient has a worse prognosis than a
patient having an ITLR value above the cut-off ITLR value. The
hazard ratio between the patient group having an ITLR below the
ITLR cut-off value and the group having an ITLR above the ITLR
cut-off value may be from around 0.2 to around 0.4, may be from
around 0.25 to around 0.36, may be around 0.25 or may be around
0.36. This means that a patient with ITLR high than the cut-off
value is 0.25-0.36 times less likely to die from breast cancer than
a patient with ITLR lower than the cut-off value. A poor prognosis
may mean that a patient has a survival probability of around 50%,
or around 49%, five years from diagnosis, or ten years from
diagnosis. A good prognosis may mean that a patient has a survival
probability of around 80% five years from diagnosis or ten years
from diagnosis.
[0124] A predetermined ITLR cut-off value may be about 0.011, or
about 0.061. The cut-off value may be from about 0.005 to about
0.070, from about 0.010 to about 0.070, from about 0.010 to about
0.012, or from about 0.050 to about 0.070. A predetermined ITLR
cut-off value for TNBC may be about 0.011, and for ovarian cancer
may be about 0.061.
[0125] ITLR was tested in two independent cohorts of TNBC and shown
to be predictive of disease-specific survival. When TNBC Cohort 1
was used as the discovery cohort an ITLR cut-off value of 0.011 was
selected (that is, patients having an ITLR below this value showed
significantly worse disease-specific survival than patients having
an ITLR above this value), and in Cohort 2 an ITLR of below 0.011
was associated with significantly worse disease-specific survival
than patients with an ITLR above 0.011 (Log-rank test p=0.0063,
FIG. 4F). Similarly, when TNBC Cohort 2 was used as the discovery
cohort an ITLR cut-off of 0.011 was selected, and in Cohort 1 an
ITLR of below 0.011 was associated with significantly worse
disease-specific survival than patients with an ITLR above 0.011
(p=0.0037, FIG. 4F).
[0126] The prognostic power of ITLR compares favourably with that
of previously published prognostic indicators. ITLR is a more
powerful prognostic indicator than the previously published
indicator "Lym" (a tumour section image-analysis based indicator of
lymphocyte abundance; Yuan et al., 2012) and several published gene
signature-based indicators (Calabro et al., Ascierto et al., Rody
et al, Ma et al., Gu-Trantien et al).
[0127] The same cut-off selection approach used to select the ITLR
cut-off was used to test the prognostic power of "Lym" (an
image-based measure of lymphocyte abundance in tumour sections) and
several gene expression signatures in TNBC and in ovarian cancer.
None of these other prognostic indicators consistently correlated
with prognosis in both Cohort 1 and Cohort 2. By contrast, ITLR
consistently stratified patients into two groups of different
clinical outcome. (See FIG. 4, FIG. 5, FIG. 14)
[0128] Compared to published gene expression signatures, ITLR was
also the only signature to show significant correlation with
disease-specific survival in multivariate Cox proportional hazards
model together with standard clinical parameters of nodal status
and tumour size in both cohorts, whichever cohort was used as the
discovery cohort (Tables 1 to 3).
[0129] Using samples from both TNBC cohorts, ITLR has a log-rank
p-value of 2.1.times.10.sup.-4 and HR 0.32 (0.17-0.58). To test the
robustness of the Cox model in determining the prognostic value of
ITLR, bootstrap analysis was used in randomly perturbed data and
the univariate and multivariate regression analysis was repeated
1,000 times. In 95.6% and 94.7% of instances, ITLR remained
significantly associated with prognosis in univariate and
multivariate analysis, respectively. Taken together, these data
support the stability and robustness of ITLR as an independent
prognostic biomarker in TNBC.
[0130] ITLR measures the ratio of intra-tumour lymphocytes to
cancer cells, thus is different to the pathological assessment
approach described in previous studies (Denkert, 2010; Loi, 2013;
Deici, 2014), where the proportion of tumour nests that were
infiltrated by lymphocytes were reported. These previous studies
agree with the results described herein, because they show that
tumour-infiltration lymphocytes are significantly correlated with
favorable outcome in TNBC. These previous approaches, like the
experiments reported herein, were based on H&E stained
pathological samples and therefore support the position that
measures of lymphocytic infiltration can be useful tool to aid
clinical decisions in TNBC.
[0131] Unlike the methods of the invention (which are based on
automated image analysis), the previous methods are based on
assessment of tumour sections by pathologists (Denkert, 2010; Loi,
2013; Deici, 2014; Salgado, 2014). The previous methods looking at
proportions of tumour nests infiltrated by lymphocytes are thus
subjective, and therefore subject to bias and variability, and
generate results relatively slowly with higher associated costs,
the previous methods are thus unsuitable for very large scale
analyses.
[0132] The approach taken in the present invention, of identifying
lymphocyte subtypes by image analysis, contrasts with previous
approaches to assessing immune infiltration in tumours. Previous
approaches using image analysis (Yuan, 2012) have only taken
account of abundance of lymphocytes in tumours, whereas approaches
that attempt to take account of spatial locations of lymphocytes
(Denkert, 2010; Loi, 2013; Deici, 2014) have used only manual
(pathologist) based processes and have relied on qualitative and
subjective assessment of cancer cell constellations and their
relationships with lymphocytes (the presence of cancer cell "nest"
and the proportion of such nests containing lymphocytes). By
contrast the present inventor has taken the approach of using image
analysis techniques to identify lymphocyte subtypes within tumours
and use the relative abundance of a subtype of lymphocytes (ITLs)
to cancer cells as an objective quantitative measure of immune
infiltration. The image analysis techniques of the present
invention are preferably automated or computer-implemented
techniques, thereby facilitating analysis of large numbers (in the
order of 100,000--preferably at least 10,000, at least 50,000, or
at least 100,000) of lymphocytes per tumour image and permitting
large-scale analyses of cohorts of patients having various types
and subtypes of cancer.
[0133] Unlike the methods of the invention, which robustly predict
clinical outcome, pathological scores of immune infiltration
(including pathological assessment of lymphocytic infiltration)
were not significantly correlated with prognosis (FIG. 13). The
pathological scores tested included PAM50 (Perou et al., 2000),
pathological assessment of lymphocytic infiltration, tumour size,
and grade. Pathological assessment of lymphocytic infiltration for
the purposes of this study was was scored as absent, mild, or
severe: Absent if there were no lymphocytes, mild if there was a
light scattering of lymphocytes, and severe if there was a
prominent lymphocytic infiltrate.
[0134] The prognostic methods of the invention, which are based on
an objective indicator of immune cell infiltration obtained by an
automated method, have several advantages over previous prognostic
methods for use in cancer. As explained above, ITLR has greater
predictive power than several previously known cancer biomarkers
and prognostic indicators and greater predictive power than
pathological scores of immune infiltration. Because ITLR is
determined using automated methods it provides an objective
measurement of immune cell infiltration in cancer (i.e. not subject
to subjective bias or human error, which causes variability in
results), it requires no pathologist scoring (and therefore no
pathologist training or following of new guidelines) and is
relatively low cost and quick to obtain, which makes it suited to
large scale analysis of cancer data. Although detection of
gene-expression signature based signatures may be automated, ITLR,
because it can conveniently be based on tumour images such as
H&E stained sections (copies of which are easily and cheaply
shared and stored long-term), is lower cost and more convenient
biomarker than gene expression signature-based biomarkers (which
require access to preserved biological samples). The image-based
ITLR outperforms several gene expression-based signatures using the
optimal cut-off selection method. In addition, considering the cost
of microarray data acquisition, the ITLR-based approaches described
herein open a new avenue for large-scale analysis on readily
available pathological samples.
TABLE-US-00001 TABLE 1 Univariate and multivariate Cox regression
results for ITLR and other signatures in two TNBC cohorts. Cohort 1
Cohort 2 HR(CI) p Conc p HR(CI) Conc ITL Uni- 0.36(0.17-0.77)
0.0063 0.601 0.25(0.09-0.69) 0.0036 0.659 ITL 0.32(0.15-0.7) 0.0042
0.668 0.15(0.05-0.43) 0.00051 0.76 Node 0.63(0.29-1.4) 0.26
4.93(1.61-15.08) 0.0052 Size 2.62(1.27-5.41) 0.0092 2.07(0.9-4.74)
0.087 Lym Uni- 0.47(0.21-1.02) 0.051 0.574 0.41(0.12-1.43) 0.15
0.575 Lym 0.48(0.22-1.05) 0.066 0.656 0.23(0.05-1.02) 0.053 0.735
Node 0.69(0.32-1.5) 0.35 4.65(1.46-14.81) 0.0092 Size
2.35(1.16-4.77) 0.018 1.66(0.65-4.25) 0.29 Calabro Uni-
0.25(0.12-0.52) 5.2 .times. 10.sup.-5 0.66 0.5(0.18-1.39) 0.18
0.587 Calabro 0.27(0.13-0.56) 3.8 .times. 10.sup.-4 0.703
0.41(0.14-1.19) 0.1 0.744 Node 0.75(0.35-1.6) 0.45 4.57(1.45-14.37)
0.0093 Size 2.26(1.07-4.76) 0.032 1.91(0.82-4.46) 0.13 Ascierto
Uni- 0.34(0.15-0.77) 0.0066 0.621 1.23(0.4-3.83) 0.72 0.51 Ascierto
0.39(0.17-0.88) 0.024 0.671 1.18(0.37-3.72) 0.78 0.735 Node
0.85(0.39-1.84) 0.68 3.6(1.21-10.7) 0.021 Size 2.06(1.02-4.16)
0.044 2.16(0.86-5.45) 0.1 IL8 Uni- 3.09(1.46-6.51) 0.0018 0.615
0(0-Inf) .sup. 0.0099 0.645 IL8 2.79(1.32-5.92) 0.0073 0.679
0(0-Inf) .sup. 1 0.808 Node 0.81(0.37-1.75) 0.59 3.14(1.06-9.34)
0.039 Size 2.23(1.08-4.63) 0.031 1.75(0.71-4.28) 0.22 CXCL13 Uni-
0.21(0.1-0.46) 1.5 .times. 10.sup.-6 0.69 0.76(0.28-2.1) 0.6 0.545
CXCL13 0.24(0.11-0.54) 4.5 .times. 10.sup.-4 0.721 0.83(0.29-2.37)
0.73 0.739 Node 0.69(0.32-1.49) 0.35 3.61(1.22-10.71) 0.021 Size
1.71(0.83-3.55) 0.15 2.12(0.86-5.22) 0.1 Shaded sections show
results from multivariate regression. Uni-: Univariate Cox
regression; HR: Hazard Ratio; CI: lower and higher 95% Confidence
Interval; Conc: Concordance; 0(0-Inf): where the Cox model failed
to converge. P-values that pass the significant threshold of 0.05
are shown in bold.
TABLE-US-00002 TABLE 2 Univariate and multivariate Cox regression
results for ITLR and other eight signatures using the optimal
cut-offs selected in Cohort 1 and validated in Cohort 2. Cohort1
Cohort2 HR(CI) p-value conc HR(CI) p-value conc Uni-ITL
0.36(0.17-0.77) 0.0063 0.601 0.25(0.09-0.69) 0.0036 0.659 Multi-ITL
0.32(0.15-0.7) 0.0042 0.668 0.15(0.05-0.43) 0.00051 0.76 Multi-node
0.63(0.29-1.4) 0.26 4.93(1.61-15.08) 0.0052 Multi-size
2.62(1.27-5.41) 0.0092 2.07(0.9-4.74) 0.087 Uni-Lym 0.47(0.21-1.02)
0.051 0.574 0.41(0.12-1.43) 0.15 0.575 Multi-Lym 0.48(0.22-1.05)
0.066 0.656 0.23(0.05-1.02) 0.053 0.735 Multi-node 0.69(0.32-1.5)
0.35 4.65(1.46-14.81) 0.0092 Multi-size 2.35(1.16-4.77) 0.018
1.66(0.65-4.25) 0.29 Uni-Calabro 0.25(0.12-0.52) 5.20E-05 0.66
0.5(0.18-1.39) 0.18 0.587 Multi-Calabro 0.27(0.13-0.56) 0.00038
0.703 0.41(0.14-1.19) 0.1 0.744 Multi-node 0.75(0.35-1.6) 0.45
4.57(1.45-14.37) 0.0093 Multi-size 2.26(1.07-4.76) 0.032
1.91(0.82-4.46) 0.13 Uni-IL8 3.09(1.46-6.51) 0.0018 0.615 0(0-Inf)
.sup. 0.0099 0.645 Multi-IL8 2.79(1.32-5.92) 0.0073 0.679 0(0-Inf)
.sup. 1 0.808 Multi-node 0.81(0.37-1.75) 0.59 3.14(1.06-9.34) 0.039
Multi-size 2.23(1.08-4.63) 0.031 1.75(0.71-4.28) 0.22 Uni-Bcell
0.6(0.25-1.48) 0.26 0.557 0.51(0.12-2.27) 0.37 0.539 Multi-Bcell
0.57(0.23-1.4) 0.22 0.655 0.48(0.11-2.17) 0.34 0.747 Multi-node
0.7(0.32-1.5) 0.35 3.77(1.27-11.2) 0.017 Multi-size 2.38(1.19-4.76)
0.014 2.07(0.86-5) 0.11 Uni-Bcell.IL8 0.52(0.24-1.11) 0.086 0.581
1.1(0.41-2.95) 0.86 0.482 Multi-Bcell.IL8 0.53(0.25-1.12) 0.097
0.648 1.22(0.42-3.53) 0.71 0.743 Multi-node 0.74(0.34-1.6) 0.44
3.76(1.24-11.41) 0.02 Multi-size 2.36(1.17-4.76) 0.016
2.21(0.88-5.54) 0.091 Uni-Ascierto 0.34(0.15-0.77) 0.0066 0.621
1.23(0.4-3.83) 0.72 0.51 Multi-Ascierto 0.39(0.17-0.88) 0.024 0.671
1.18(0.37-3.72) 0.78 0.735 Multi-node 0.85(0.39-1.84) 0.68
3.6(1.21-10.7) 0.021 Multi-size 2.06(1.02-4.16) 0.044
2.16(0.86-5.45) 0.1 Uni-CXCR3 0.3(0.14-0.64) 9.00E-04 0.618
0.82(0.3-2.25) 0.7 0.535 Multi-CXCR3 0.31(0.15-0.66) 0.0026 0.683
0.79(0.25-2.45) 0.68 0.73 Multi-node 0.86(0.39-1.87) 0.7
3.81(1.24-11.72) 0.02 Multi-size 2.24(1.13-4.44) 0.02
2.07(0.82-5.18) 0.12 Uni-CXCL13 0.21(0.1-0.46) 1.50E-05 0.69
0.76(0.28-2.1) 0.6 0.545 Multi-CXCL13 0.24(0.11-0.54) 0.00045 0.721
0.83(0.29-2.37) 0.73 0.739 Multi-node 0.69(0.32-1.49) 0.35
3.61(1.22-10.71) 0.021 Multi-size 1.71(0.83-3.55) 0.15
2.12(0.86-5.22) 0.1 Uni-: Univariate Cox regression; Multi-:
Multivariate Cox regression; HR: Hazard Ratio; CI: lower and higher
95% Confidence Interval; Conc: Concordance; Inf: Cox model failed
to converge.
TABLE-US-00003 TABLE 3 Univariate and multivariate Cox regression
results for ITLR and other eight signatures using the optimal
cut-offs selected in Cohort 2 and validated in Cohort 1. Cohort1
Cohort2 HR(CI) p-value conc HR(CI) p-value conc Uni-ITL
0.45(0.21-0.96) 0.033 0.587 0.26(0.1-0.71) 0.0048 0.656 Multi-ITL
0.38(0.17-0.84) 0.016 0.654 0.16(0.05-0.48) 0.001 0.76 Multi-node
0.62(0.28-1.37) 0.23 4.64(1.52-14.15) 0.007 Multi-size
2.62(1.27-5.39) 0.0088 2.07(0.89-4.83) 0.091 Uni-Lym
0.91(0.43-1.91) 0.8 0.524 0.35(0.13-0.98) 0.038 0.63 Multi-Lym
0.92(0.43-1.95) 0.82 0.627 0.29(0.1-0.85) 0.024 0.778 Multi-node
0.72(0.33-1.58) 0.41 3.82(1.29-11.38) 0.016 Multi-size
2.33(1.18-4.64) 0.015 1.85(0.73-4.73) 0.2 Uni-Calabro
0.53(0.24-1.2) 0.12 0.578 0(0-Inf) .sup. 0.04 0.608 Multi-Calabro
0.56(0.24-1.28) 0.17 0.667 0(0-Inf) .sup. 1 0.799 Multi-node
0.67(0.31-1.46) 0.31 4.12(1.39-12.23) 0.011 Multi-size
2.17(1.11-4.24) 0.023 1.91(0.8-4.55) 0.14 Uni-IL8 1.76(0.86-3.6)
0.12 0.575 0.18(0.05-0.63) 0.0026 0.692 Multi-IL8 1.74(0.84-3.59)
0.14 0.65 0.21(0.06-0.77) 0.018 0.795 Multi-node 0.8(0.37-1.73)
0.57 3.76(1.25-11.34) 0.019 Multi-size 2.29(1.17-4.51) 0.016
1.87(0.68-5.17) 0.23 Uni-Bcell 0.74(0.35-1.59) 0.44 0.541
0.41(0.09-1.82) 0.23 0.557 Multi-Bcell 0.75(0.35-1.6) 0.45 0.63
0.33(0.07-1.5) 0.15 0.763 Multi-node 0.72(0.33-1.55) 0.39
4.24(1.41-12.76) 0.01 Multi-size 2.33(1.17-4.61) 0.016
1.94(0.81-4.67) 0.14 Uni-Bcell.IL8 0.82(0.35-1.92) 0.65 0.511
0.45(0.16-1.31) 0.13 0.599 Multi-Bcell.IL8 0.75(0.32-1.76) 0.51
0.629 0.59(0.2-1.76) 0.34 0.753 Multi-node 0.73(0.34-1.59) 0.43
3.28(1.08-9.94) 0.036 Multi-size 2.4(1.2-4.83) 0.014 2.18(0.85-5.6)
0.11 Uni-Ascierto 0.83(0.39-1.78) 0.64 0.52 2.26(0.82-6.23) 0.11
0.63 Multi-Ascierto 1(0.46-2.16) 0.99 0.619 2.6(0.87-7.82) 0.089
0.773 Multi-node 0.73(0.34-1.59) 0.43 3.22(1.08-9.61) 0.036
Multi-size 2.33(1.16-4.66) 0.017 2.46(0.91-6.62) 0.075 Uni-CXCR3
0.56(0.24-1.29) 0.17 0.569 0(0-Inf) .sup. 0.029 0.618 Multi-CXCR3
0.62(0.26-1.48) 0.28 0.658 0(0-Inf) .sup. 1 0.805 Multi-node
0.68(0.32-1.49) 0.34 4.17(1.41-12.38) 0.01 Multi-size
2.16(1.09-4.28) 0.028 1.89(0.8-4.46) 0.15 Uni-CXCL13
0.35(0.17-0.75) 0.0043 0.605 2.21(0.71-6.86) 0.16 0.595
Multi-CXCL13 0.38(0.18-0.79) 0.01 0.663 3.38(0.92-12.45) 0.067
0.773 Multi-node 0.69(0.32-1.49) 0.35 3.71(1.24-11.1) 0.019
Multi-size 2.29(1.11-4.72) 0.026 2.71(0.94-7.8) 0.064 Uni-:
Univariate Cox regression; Multi-; Multivariate Cox regression; HR:
Hazard Ratio; CI: lower and higher 95% Confidence Interval; Conc:
Concordance.
[0135] ITLR as an unbiased assessment of immune infiltration can
facilitate the discovery of molecular correlates with this
clinically important phenomenon. While the expression of many
immune-related genes in tumours was significantly associated with
ITLR, it is unclear whether these genes are expressed on cancer
cells or lymphocytes. This is because the microarray data were
obtained using whole-tumour materials without micro-dissection.
[0136] The data herein show that the RNA expression of cytotoxic
T-lymphocyte-associated protein 4 (CTLA4), a receptor of the
immunoglobulin family and the target of ipilimumab, was
significantly associated with ITLR as well as longer disease
specific survival in TNBC. This is consistent with the recent
observation in non-small cell lung cancers that over-expression of
CTLA4 is associated with reduced death rate (Salvi, 2012). CTLA4 is
expressed in tumour cells in different cancer types (Contardi,
2005). In breast cancer it is expressed in both tumour cells and T
cells, and an inverse correlation between CTLA4 expression and
clinical outcome (i.e. high CTLA4 expression associated with poor
clinical outcome) has been previously reported in 60 patients with
different breast cancer subtypes (Mao, 2010), which is in contrast
with the data herein from TNBC (see below), and which thus
highlights the novel molecular insights into cancer yielded by
ITLR. A recent study showed that in situ mRNA expression of another
receptor of the immunoglobulin superfamily, PDL1, is associated
with increased immune infiltration and favourable recurrence free
survival across different breast cancer subtypes (Schalper,
2014).
[0137] Taken together, the data herein support the potential of
CTLA4-targeted therapies in TNBC. CTLA4 is a negative regulator of
T cells, and therefore its expression reduces T cell-mediated
killing of cancer cells. The data herein show a positive
association between CLTA4 expression and ITLR, consistent with ITL
expression of CTLA4. The expression of CTLA4 in ITLs may explain
why in many tumours cancer cells were not eliminated even in the
presence of high numbers of ITLs. The use of CTLA4 antagonists to
inhibit immune tolerance to cancer and to activate ITLs may be an
effective treatment strategy for TNBC.
[0138] Unsupervised clustering with Gaussian Mixture modelling for
CTLA4 expression in all 1,980 METABRIC tumours revealed two
clusters, one with high and one with low level of expression of
CTLA4 (FIG. 15 A). Using this clustering definition for TNBC
tumours we found that TNBC patients with higher level of CTLA4
expression have significantly better disease-specific survival than
patients with lower level of CTLA4 expression (p=0.018, HR=0.61,
CI=0.41-0.92, FIG. 15 B).
[0139] The gene module analysis also revealed several tightly
connected, functionally related modules. For example, one module
contains APOBEC3G (Apolipoprotein B MRNA Editing Enzyme, Catalytic
Polypeptide-Like 3G), which is known to play important roles in
adaptive and innate immunity and has been investigated extensively
in viral infection (Mangeat, 2003) but its role in breast cancer
has not been investigated in detail. It is a member of the
apolipoprotein B mRNA-editing enzyme, catalytic polypeptide-like
editing complex family together with APOBEC3B, which was found to
be a source of mutagenesis in many major cancer types including
breast cancer (Kuong, 2013). In the TNBC samples studied herein,
APOBEC3G expression is significantly correlated with favourable
prognosis (log-rank p=0.02) but not other APOBEC members including
APOBEC3B (p=0.29). APOBEC3G is primarily expressed in CD4+T
lymphocytes, macrophages, and dendritic cells (Monajemi, 2012). The
present data revealed strong association between APOBEC3G and
natural killer cell gene NKG7 and interleukins in this module and
support the importance of APOBEC3G in TNBC.
[0140] The associations between ITLR and immune-relevant genes,
pathways and modules support the validity of ITLR as a measure of
lymphocytic infiltration and reveal co-regulations of key immune
genes.
[0141] An aspect of the present invention provides a method of
determining a prognosis in a triple negative breast cancer patient,
the method comprising, determining the level of expression of
APOBEC3G in a tumour sample obtained from the patient, wherein
increased expression and/or APOBEC3G expression indicates a
positive prognosis.
[0142] An aspect of the present invention provides a method of
determining a prognosis in a triple negative breast cancer patient,
the method comprising, determining the level of expression of CTLA4
in a tumour sample obtained from the patient, wherein increased
CTLA4 expression indicates a positive prognosis. Increased CTLA4
expression may be CTLA4 expression above the middle (50) percentile
for TNBC, or may be CLTA4 expression above the (25) percentile for
TNBC. Increased CTLA4 expression may be CTLA4 expression that is
high relative to one or more "housekeeping" genes such as
glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
[0143] A method of determining a prognosis based on ITLR as
described herein may further comprise a step of measuring CTLA4
expression in a tumour obtained from the cancer patient. The cancer
patient may be a TNBC patient. The step of measuring CTLA4
expression may involve nucleic acid hybridisation (e.g.
microarray-based analysis) or immunohistochemical techniques. In
such methods, the combination of an ITLR above a predetermined
cut-off value and increased CTLA4 expression indicates a positive
prognosis. The predetermined cut-off value for ITLR may be about
0.03 or about about 0.032.
[0144] ITLR is an objective quantitative indicator of lymphocytic
infiltration in tumours. This quantitative measurement of immune
infiltration is useful in guiding treatment decisions in
cancer.
[0145] Accordingly, an aspect of the invention provides a method of
using ITLR in predicting whether or not a cancer patient will
respond to a therapy.
[0146] Such a method may be a method for predicting whether a
cancer patient will respond to a therapeutic regime, the method
comprising measuring immune infiltration in a tumour from the
cancer patient according to a method described herein, wherein an
ITLR above a predetermined ITLR cut-off value indicates that the
patient is likely to respond to the therapeutic regime.
[0147] ITLR is useful in informing treatment decisions for cancer
patients. Accordingly, an aspect of the present invention provides
a method of treating a cancer patient, wherein the ITLR of the
tumour has been determined to be either below, or above, a
predetermined cut-off value. The cancer patient may be an
individual from whom an image of a tumour has been obtained. The
method may comprise determining the ITLR in a tumour from the
patient. The method of treatment may comprise administration of a
therapeutic regime.
[0148] The therapeutic regime may be radiotherapy or chemotherapy
or any combination of these. A therapeutic regime comprising
chemotherapy may comprise anthracyline-based chemotherapy. A
therapeutic regime may comprise administration of a therapeutic
agent. Accordingly, an aspect of the present invention provides a
therapeutic agent for use in treating cancer in a cancer patient,
wherein a prognosis for the cancer patient has been obtained using
a method as disclosed herein.
[0149] ITLR provides information for predicting a long-term
prognosis and for informing patient treatment decisions. Thus if a
patient has a low ITLR and is likely to have a poor prognosis, this
patient may be treated more intensively (e.g. more rounds of
chemotherapy) than a patient having a high ITLR.
[0150] Accordingly, an aspect of the present invention provides a
method of treating cancer in a cancer patient according to a
therapeutic regime, the method comprising analysing a tumour image
from the cancer patient according to a method described herein, and
treating the cancer patient according to the therapeutic regime
depending on whether the ITLR is below or above a predetermined
cut-off value.
[0151] ITLR combined with CTLA4 expression provides further
prognostic information. Relatively high CTLA4 expression may be
associated with a high ITLR, and inhibition of CTLA4 may activate T
cells to kill cancer cells. Thus, for a patient having an ITLR
above a predetermined cut-off value and having increased CTLA4
expression the therapeutic regime may comprise administration of a
CTLA4 antagonist. The CTLA4 antagonist may be an antibody, for
example ipilimumab.
[0152] An aspect of the present invention provides a CTLA4
antagonist for use in a method of treatment of cancer, wherein a
tumour from the patient has been determined to have a high ITLR. An
aspect of the present invention provides a CTLA4 antagonist for use
in a method of treatment of cancer, wherein a tumour from the
patient has been determined to have an ITLR above a predetermined
ITLR cut-off value. The cancer may be a specific type or subtype of
cancer and the predetermined ITLR cut-off value may be the cut-off
value determined for a cohort of patients having that cancer type
or subtype. The cancer subtype may be breast cancer. The CTLA4
antagonist may be an antibody, which may be an anti-CTLA4 antibody.
The anti-CTLA4 antibody may be ipilimumab (also known as MDX-010
and MDX-101). The cancer patient may be a TNBC patient and the
therapy may be ipilimumab.
[0153] The prognostic and therapeutic methods described herein may
further comprise surgically resecting a tumour from a cancer
patient, measuring immune infiltration in the tumour according to a
method described herein, and determining a prognosis and/or
treating the cancer patient according to a therapeutic regime based
on the ITLR of the tumour. A surgically resected tumour is a
surgically removed tumour. The method of measuring immune
infiltration may use a whole tumour section.
[0154] An aspect of the present invention provides a method of
determining the efficacy of a therapeutic regime. The method may
comprise determining the ITLR of a tumour biopsy obtained from a
patient before undergoing the therapeutic regime, determining the
ITLR of a tumour biopsy obtained from the patient after undergoing
the therapeutic agent, and associating an increased ITLR with
therapeutic efficacy (i.e. a therapeutic effect).
[0155] The methods of analyzing tumours according to the invention
may be modified to yield further information on lymphocyte subtypes
and their relevance in cancer. Lymphocytes in tumours are known to
encompass diverse subclasses including helper T cells, regulatory T
cells, natural killer cells and B cells with sophisticated
implications for treatment response (Fridman, 2012; Gu-Trantien,
2013; Andre, 2013). Immunohistochemistry analysis of tumour
sections with immune cell markers may be performed, for which
automated immunohistochemistry image analysis and statistical
modelling methods could be developed to discern interactions
between cancer and anti-/pro-tumoural immune response.
[0156] In the context of the methods and therapeutic agents
described herein, a pathological section may be a tumour section. A
tumour section may be a whole-tumour section. A whole-tumour
section is typically a section cut from a surgically resected
tumour, thus representing the characteristics of the whole tumour.
Thus, a whole-tumour section may be a surgically resected section.
A pathological section may be a biopsy obtained from a tumour. The
pathological section is preferably stained. Staining facilitates
morphological analysis of tumour sections by colouring cells,
subcellular structures and organelles. Any type of staining may be
used, provided that the staining facilitates morphological
analysis. The pathological section may be stained with hematoxylin
and eosin (H&E). H&E stain is the most commonly used stain
in histopathology for medical diagnosis, particularly for the
analysis of biopsy sections of suspected cancers by pathologists.
Thus H&E stained pathological sections are usually readily
available as part of large data sets collated for the study of
cancer. The applicability of the present methods to H&E stained
pathological sections makes them particularly adaptable for use in
analysing data sets from many types and subtypes of cancer to
determine the prognostic value of ITLR and to determine cut-off
ITLR values for use in the methods described herein.
[0157] Reference herein to the ITLR of a tumour also refers to the
ITLR of a pathological section, tumour section, or tumour
image.
[0158] Reference herein to an ITLR value being below a cut-off
value may also refer to an ITLR value being equal to or below a
cut-off value.
[0159] In the present context the term tumour image refers to an
image of a tumour from a patient. A tumour image may be an image of
a pathological section or tumour section. In the present disclosure
a patient may be referred to as having an ITLR (e.g. an ITLR below
a predetermined cut-off value), meaning that an image of a tumour
from that patient has been determined to have an ITLR. The tumour
image may be of a section of a surgically resected tumour, or may
be of a biopsy of a tumour.
[0160] In the present context the ratio of intra-tumour lymphocytes
to cancer cells (ITLR) is the ratio in the pathological section, in
a tumour, in a biopsy from the tumour, in a tumour section, or in
an image of the tumour, tumour section or pathological section. The
term ITLR may also be attributed to a patient. A patient having an
ITLR of a particular value refers to a patient from whom a
pathological section has an ITLR of a particular value.
[0161] The terms "lymphocytic infiltration" and "immune
infiltration" are used interchangeably herein.
[0162] In the present context "automated" refers to processes that
operate independent of external (human) control or input. In the
present context an automated process may be a computer-implemented
process. The methods of the present invention may be automated
methods. The methods of the present invention may be entirely
automated methods, that is, they may operate independently of human
control or input in their entirety. The methods of the present
invention may comprise a step of identifying lymphocyte and cancer
cells by automated image analysis. The methods of the present
invention may be performed on a tumour image in which lymphocytes
and cancer cells have been identified by automated image
analysis.
[0163] The methods of the present invention are performed on
pathological sections, such as tumour sections. The methods of the
present invention are therefore ex vivo methods, that is, the
methods of the present invention are not practiced on the human
body.
[0164] A cancer patient in the context of the present invention is
an individual having cancer or having been diagnosed with cancer.
Reference to cancer may be reference to a particular type or
subtype of cancer. The cancer patient may have undergone
anthracyline-based chemotherapy, immunotherapy, or a combination
therapy comprising anthracyline-based chemotherapy and
immunotherapy. The cancer patient may have breast cancer,
colorectal cancer, melanoma or non-small cell lung cancer. The
cancer patient may have the subtype of breast cancer known as
triple negative breast cancer. Triple negative breast cancer may be
defined as a breast cancer that is negative for estrogen receptors
(ER) and HER2. (TNBC is sometimes defined as breast cancer that is
negative for estrogen receptors (ER), HER2 and progesterone
receptors (PR), but since cancer that are negative for ER are
typically also negative for PR, in the present context TNBC is
defined as breast cancer that is negative for ER and HER2.
[0165] In the context of the present invention reference to the
treatment of a cancer patient refers to treatment of cancer in a
patient.
[0166] The cancer patient may have, or the cancer type or subtype
may be selected from, breast cancer (including breast ductal
carcinoma breast and breast lobular carcinoma), central nervous
system cancer (including glioblastoma multiforme and lower grade
glioma), endocrine cancer (including adrenocortical carcinoma,
papillary thyroid carcinoma, paraganglioma & pheochromocytoma),
gastrointestinal cancer (including Cholangiocarcinoma, Colorectal
Adenocarcinoma, Liver Hepatocellular Carcinoma, Pancreatic Ductal
Adenocarcinoma, Stomach-Esophageal Cancer), gynecologic cancer
(including Cervical Cancer, Ovarian Serous Cystadenocarcinoma,
Uterine Carcinosarcoma, Uterine Corpus Endometrial Carcinoma), head
and neck cancer (including Head and Neck Squamous Cell Carcinoma,
Uveal Melanoma), hematologic cancer (including Acute Myeloid
Leukemia, and Acute Myeloid Leukemia), skin cancer (including
Cutaneous Melanoma), soft tissue cancer (including Sarcoma),
thoracic cancer (including Lung Adenocarcinoma, Lung Squamous Cell
Carcinoma, Mesothelioma) and urologic cancer (including Chromophobe
Renal Cell Carcinoma, Clear Cell Kidney Carcinoma, Papillary Kidney
Carcinoma, Prostate Adenocarcinoma, Testicular Germ Cell Cancer,
Urothelial Bladder Carcinoma). Each of these cancers is the subject
of study as part of The Cancer Genome Atlas project.
[0167] In the present context the term "immune signatures" is used
to encompass all biomarkers related to immune responses and
includes the gene expression signatures studied herein as well as
other biomarkers including the "Lym" biomarker (Yuan et al. 2012)
and ITLR.
[0168] Each and every compatible combination of the embodiments
described above is explicitly disclosed herein, as if each and
every combination was individually and explicitly recited.
[0169] Various further aspects and embodiments of the present
invention will be apparent to those skilled in the art in view of
the present disclosure.
[0170] "and/or" where used herein is to be taken as specific
disclosure of each of the two specified features or components with
or without the other. For example "A and/or B" is to be taken as
specific disclosure of each of (i) A, (ii) B and (iii) A and B,
just as if each is set out individually herein.
[0171] Unless context dictates otherwise, the descriptions and
definitions of the features set out above are not limited to any
particular aspect or embodiment of the invention and apply equally
to all aspects and embodiments which are described.
Experimental
[0172] Methods
[0173] Breast Cancer Studies
[0174] Clinical Samples
[0175] The complete set of METABRIC (Curtis et al.) samples
contains 1,980 primary frozen breast tumours from five contributing
hospitals. Among these, 1,026 of the 1,047 tumours from three
hospitals have H&E sections without severe artefacts, whist all
the H&E samples from the other two hospitals are highly
fragmented due to long-term frozen storage. Therefore we only
considered the 1,026 tumours for this study (long-term follow up
median 68.3 months). On average three tumour sections were obtained
at different locations of each primary tumour and placed onto the
same slide (Yuan et al., 2012). Tumour materials sandwiched between
these sections were sectioned, mixed and used for molecular
profiling, thereby maximising the biological relevance of multiple
data types being generated. Further details on experimental
procedure, staining and molecular profiling protocols can be found
in Yuan et al 2012. Gene expression data for the same set of
tumours were profiled using the Illumina HT-12 platform. ER status
was determined based on the bimodal distribution of ESR1 expression
microarray data, and Her2 amplification status based on microarray
SNP6 data from the same tumours. In total, there were 181
ER-negative, Her2-negative samples and these were defined as triple
negative/TNBC. Samples from two of the three hospitals were merged
to form Cohort 1 (89 samples) and samples from the other hospital
were merged to form Cohort 2 (92 samples) in order to obtain a
similar population size in each cohort. Immune infiltration was
scored for 112 of the 181 samples by the pathologists in the
METABRIC consortium into three categories: absent, mild and severe.
Absent if there were no lymphocytes, mild if there was a light
scattering of lymphocytes, and severe if there was a prominent
lymphocytic infiltrate. The pathological scores of immune
infiltration were not significantly correlated with prognosis (FIG.
13).
[0176] H&E Image Analysis
[0177] The accuracy of the automated image analysis tool for
H&E breast tumour frozen section images had previously been
validated based on pathological tumour scores and cell-by-cell
evaluation (Yuan et al., Natrajan et al.). For METABRIC samples,
this tool achieved 90% cross-validation accuracy for cell
classification and high correlation with pathological scores of
cell proportions (cor=0.98) (Yuan et al.). This tool was used to
classify all cell nuclei in 181 TNBC whole-tumour sections,
resulting in an average of 81,810 (standard deviation 80,330)
cancer cells, 15,500 (25,133) lymphocytes, and 14,090 (14,180)
stromal cells for each image. Lymphocytes have a typical morphology
of small, round and homogeneously basophilic nuclei, thus can be
reliably differentiated from other cell types in cancer. Since this
analysis is based on nuclear morphology only in the H&Es, the
identified lymphocytes are likely to be a mixture of immune cell
types including T- and B-lymphocytes.
[0178] Modelling the Spatial Heterogeneity of Cancer-Immune
Interaction
[0179] Let x=x.sub.1, x.sub.2, . . . x.sub.n be the spatial
locations of n cancer cells and y=y.sub.1, y.sub.2, . . . ,
y.sub.rn be the spatial locations of m immune cells in a tumour
image (e.g. an H&E tumour section image). Using a quartic
kernel function K one can establish a kernel density estimate over
the whole tumour image:
f ( x ) = i n K ( x - x i ) h , ##EQU00001##
where h is the bandwidth parameter for K. h was optimised using the
Minimum Square Error criteria (Berman et al.) in 10 randomly
sampled images. Thus, the spatial proximity to cancer for an immune
cell i is s.sub.i=f(y.sub.i). We can then identify lymphocyte
classes based on s, s=s.sub.1, s.sub.2, . . . , s.sub.m, using
unsupervised Gaussian Mixture Clustering (McLachlan, 2000). This
method aims to identify multiple components/clusters within the
data with probabilities that quantify the uncertainty of
observations belonging to the clusters.
p ( s ) k = 1 K w k G ( s | .mu. k , .sigma. k ) , ##EQU00002##
where K is the number of clusters, .mu..sub.k and .sigma..sub.k are
the mean and variance that define the probabilistic density
function G for the kth component, and w.sub.k is the weight of a
component k. These parameters were estimated by
Expectation-Maximization (Dempster, 1977). Selection of models with
different numbers of clusters can be done using statistical
criteria, one of the most common being the Bayesian Information
Criterion (Schwartz, 1978). It can be used in conjunction with
mixture model clustering to select the best number of clusters
K:
BIC=2L(p(s))+d log(m)
where L( ) is the maximum log likelihood function and d is the
number of free parameters to be estimated. Effectively, the BIC
criterion aims to evaluate modelling error as well as model
complexity. The higher the value of BIC and better the solution is
considered to be. To perform clustering, 100,000 immune cells were
randomly sampled. Their spatial proximity to cancer data s were
used for clustering with a range of different K, K=1-5. This was
repeated 200 times, 97% of which the solution with three clusters
was considered the optimal by BIC. Mean .mu..sub.k of the clusters
are consistent (median: 0.011, 0.06, 0.13; standard
deviation/SD:
[0180] 0.002, 0.0047, 0.0045). Subsequently, we classified all
lymphocytes in all tumour samples based on these clusters. We used
the ratio of the number of intra-tumour lymphocytes and the number
of cancer cell as the final measurement of intra-tumour immune
infiltration:
ITLR = N Intra - Tumour Lymphocyte N cancer ##EQU00003##
[0181] Image Analysis and Modelling Spatial Heterogeneity in More
Detail
[0182] Image Data
[0183] CRImage processes a H&E slide by first dividing it into
2,000 pixels by 2,000 pixels sub-images and identifying cells in
these sub-images. Therefore the cell locations for these sub-images
need to be combined. We provide combined cell identifies and
spatial locations for all 181 TNBC whole-section H&E sections
as R data files in a `CellPosAndMask` folder. These files are named
by their image ID. Each file contain the x, y and class columns
storing x y coordinates as well as the class of each cell in the
large H&E slide. There is also a `mask` binary matrix to denote
the tissue area. The resolution of this image is 5 .mu.m per
pixel.
[0184] Identify the Optimal Bandwidth Parameter for Computing
Kernel Density
[0185] By sampling 10 random samples, the Mean Square Error is
computed over a range of different bandwidths h for computing
cancer density.
TABLE-US-00004 library(splancs) MSE <- NULL set.seed(10) ffs
<- sample(dir(`./data/CellPosAndMask/`), 10) for (ff in ffs){
res <- try(load(paste(`./data/CellPosAndMask/`, ff, sep=``)))
CellPos[,1] <- as.character(CellPos[,1]) CellPos[,2] <-
as.numeric(CellPos[,2]) CellPos[,3] <- as.numeric(CellPos[,3])
CellPos <- CellPos[rowSums(is.na(CellPos))==0, ] CellPos[,3]
<- ncol(Mask) - CellPos[,3] +1 CellPos[,3][ CellPos[,3] >
ncol(Mask)] <- ncol(Mask) CellPos <-
CellPos[CellPos[,1]!=`a`,] cell.c <-
data.frame(x=as.numeric(CellPos[CellPos[,1]==`c`,2]),
y=as.numeric(CellPos[CellPos[,1]==`c`,3])) cv <-
mse2d(as.points(cell.c), poly=cbind(c(0, 0, nrow(Mask),
nrow(Mask)), c(0, ncol(Mask), ncol(Mask), 0)), nsmse=40, range=10)
MSE <- rbind(MSE, cv$mse) } save(cv, MSE,
file=`./data/BandwidthSelection.rdata`) h=5 was chosen as the
optimal bandwidth for lower variability of Mean Square Error.
[0186] Generate Spatial Proximity to Cancer for Each Lymphocyte
[0187] Now, spatial scores can be generated given the cell position
data using the following getITL function. getITL function uses the
cell position files to infer a cancer density map using the
bandwidth selected above.
TABLE-US-00005 getITL <- function(ff, ...){ require(EBImage)
require(splancs) res <- try(load(paste(`./data/CellPosAndMask/`,
ff, `.rdata`, sep=''))) if (class(res)!=`try-error`){ CellPos[,1]
<- as.character(CellPos[,1]) CellPos[,2] <-
as.numeric(CellPos[,2]) CellPos[,3] <- as.numeric(CellPos[,3])
CellPos <- CellPos[rowSums(is.na(CellPos))==0, ] CellPos[,3]
<- ncol(Mask) - CellPos[,3] +1 CellPos[,3][ CellPos[,3] >
ncol(Mask)] <- ncol(Mask) cell.c <-
data.frame(x=as.numeric(CellPos[CellPos[,1]==`c`,2]),
y=as.numeric(CellPos[CellPos[,1]==`c`,3])) res <-
kernel2d(as.points(cell.c), poly=cbind(c(0, 0, nrow(Mask),
nrow(Mask)), c(0, ncol(Mask), ncol(Mask), 0)), h0=h,
nx=dim(Mask)[1], ny=dim(Mask)[2]) cell.l <-
data.frame(x=as.numeric(CellPos[CellPos[,1]==`l`,2]),
y=as.numeric(CellPos[CellPos[,1]==`l`, 3])) z.l <-
unlist(sapply(1:length(cell.l$x), function(x) res$z[cell.l$x[x],
cell.l$y[x]])) } z.l }
[0188] Using this function, measurements for each lymphocyte for
each tumour can then be generated.
TABLE-US-00006 itl <- list( ) files <- trait$file for (ff in
files) itl <- c(itl, list(try(getITL(ff, h=5, w=3, cex=.5,
ifPlot=F)))) names(itl) <- files save(itl,
file=`./data/ITL.rdata`)
[0189] By default getITL function uses the cut-offs (threshold
values) of 0.10507473 and 0.03662728 to determine intra-tumour
(ITL), adjacent to tumour (ATL), and distal-to-tumour lymphocytes
(DTL). We will now describe how these cut-offs were selected.
[0190] Identify Sub-Populations of Lymphocyte by Unsupervised
Learning
[0191] Gaussian mixture clustering and BIC implemented in the R
package mclust were used for the discovery of lymphocyte
sub-populations. 100,000 lymphocytes were randomly sampled from the
itl object and then clustered.
TABLE-US-00007 library(mclust) load(file=`./data/ITL.rdata`)
set.seed(11) x <- sample(as.numeric(unlist(itl)), 100000) res
<- Mclust(x, G=1:5)
[0192] The sampling process was repeated to generate clusters 200
times, and evaluated output from Mclust and mclustBIC. BIC values
for these 200 runs are obtained. The three-cluster solution k=3
remains optimal in 97% of the time, and k=5 was chosen 3% of the
times. The median of cluster means when there are three clusters
are 0.0114, 0.0603 and 0.1322 with standard deviation 0.002 and
0.0047 and 0.0045, respectively.
[0193] Therefore, the clustering result of lymphocytes from
randomly sampled data is stable. Since the clustering is stable,
cut-offs were taken at the maximum value of the first and second
clusters from one of the sampling runs as our cut-offs for
determining lymphocyte classification for the remaining
samples.
[0194] Generating ITLR, ATLR, and DTLR
[0195] Subsequently, the cut-offs can be used to classify every
lymphocyte based on their data stored in the R object itl. mat.I is
a matrix with columns of `Distal`, `Adjacent`, `Intra` denoting the
number of lymphocytes in each class for a tumour.
TABLE-US-00008 th=c(0.03662728, 0.10507473) mat.l <- NULL for (i
in 1:length(itl)){ z.l <- itl[[i]] cl <- rep(1,length(z.l))
cl[z.l>th[1] & z.l<th[2]] <- 2 cl[z.l>=th[2]] <-
3 mat.l <- rbind(mat.l, c(sum(cl==1), sum(cl==2), sum(cl==3))) }
colnames(mat.l) <- c(`Distal`, `Adjacent`, `Intra`)
rownames(mat.l) <- names(itl)
[0196] The Intra column of mat.I is the number of intra-tumour
lymphocytes. This divided by the number of cancer cells
(trait$nTumour) is the ITLR measurement
[0197] Measuring Cell Distances and Spatial Arrangement
[0198] To identify physical properties of ITLs, ATLs and DTLs that
differentiate them, in 10,000 lymphocytes randomly sampled from 20
tumours, we identified the 5 nearest cancer cells and the centroid
of the convex hull region formed by these cancer cells. For each
lymphocyte, the distance from the lymphocyte to the nearest cancer
cell was computed (d.sub.min), and the distance to the centroid of
cancer convex hull was computed (d.sub.centroid). Centroid of a
convex hull region was calculated as the mean positions of the
subset of points that define the convex hull. Differences among
lymphocyte classes in terms of d.sub.centroid and and
d.sub.centroid were tested with student's t-test.
[0199] Other Immune Signatures in Comparison
[0200] Lymphocyte abundance based on image analysis result was
calculated as:
lym = N lymphocyte N cancer ##EQU00004##
[0201] The gene expression signatures were calculated as described
in the referred papers.
[0202] ITLR Gene Modules
[0203] Hierarchical clustering was used to identify highly
correlated gene modules by clustering the correlation matrix of all
ITL-associated genes into 100 clusters. Modules were selected from
these clusters based on average absolute Pearson correlation
exceeding 0.75 and cluster size exceeding five.
[0204] Comparing ITLR and ITLR-Associated Genes
[0205] To test if ITLR has additional value to ITLR-associated
genes, we performed multivariate Cox regression analysis with ITLR
paired with expression profile of an ITLR gene. This was performed
for all of the top 100 ITLR-associated genes ranked by correlation.
ITLR was dichotomised using the threshold reported in the paper,
and gene expression was dichotomised into two equal-size group or
three groups (25 lower, 50 middle and 25 upper percentiles). Tables
with Hazard ratio, log-rank p-value and 95% interval were produced.
In both analysis with two and three patient groups according to
gene expression data, p-values of ITLR were consistently higher
than the p-values of gene expression profiles, as well as being
higher than significance level of 0.05 (-log(p) 2.99).
[0206] Other Statistical Methods
[0207] Monotone trend between ITLR and clinical parameters was
tested using the Jonckheere-Terpstra trend test (Jonckheere).
Survival analysis was performed with breast cancer-specific 10-year
survival data. The Kaplan-Meier estimator was used for patient
stratification and log-rank test was used for testing difference
among groups. Cox proportional hazards regression model was fitted
to the survival data and hazard ratios and 95% confidence intervals
were computed to determine the correlation with disease-specific
survival, where the log-rank test with p<0.05 was considered
significant. Correlation between ITLR and gene expression was
computed with Pearson correlation and q-values computed using False
Discovery Rate (FDR) correction using 25% of the data for fitting
the null model. Cut-offs for dichotomizing immune signatures were
optimised stepwise from 20 to 80 percentiles at an interval of 1.5.
The cut-offs that displayed the highest prognostic significance
with log-rank test were selected. For consistency test in FIG. 5F,
each signature was centred at 0 and scaled to standard deviation 1.
Optimal cut-offs were also mapped to the new data before
comparison. MSigDB gene set version 4.0 (Subramanian et al.) was
used in conjunction with a hypergeometric test for enrichment
analysis.
[0208] Ovarian Cancer Studies
[0209] Samples were obtained from a UK-China collaborative study
which aims to study the clinical implications of immune
infiltration in a set of 91 ovarian cancer patients with metastatic
disease. H&E-stained slides for the primary tumours were
obtained, scanned, and subjected to image analysis using CRImage.
Cells in these images were classified into cancer, lymphocyte, and
stromal cell categories. Once the spatial locations of these cells
were obtained from image analysis, kernel density of cancer was
computed for each image, and lymphocyte-to-cancer measurements were
obtained for each lymphocyte. The measurements were subjected to
clustering and two clusters were found, i.e. intra-tumour and
non-intra tumour lymphocytes. ITLR as the ratio of intra-tumour
lymphocytes to cancer cells was calculated for each patient. The
29% of patients with ITLR lower than a cut-off of 0.06085726 have
significantly worst overall survival than patients with ITLR higher
than the cut-off (10-year OS log-rank test p=0.024, HR=0.51,
CI=0.28-0.92; 5-year OS p=0.045, HR=0.54, CI=0.29-0.99; Figs
Ovarian). Overall survival was defined using death as event
regardless of the cause, as this information was unavailable.
[0210] Tumours were staged according to the 1988 FIGO staging
system (Prat 2013). Lymphocytic infiltration was assessed in five
high-power fields, each field is scored as absent, mild, or severe:
Absent if there were no lymphocytes, mild if there was a light
scattering of lymphocytes, and severe if there was a prominent
lymphocytic infiltrate. Median of field-based scores was taken as
the score for a tumour.
[0211] Results
[0212] Statistical Modelling of the Spatial Heterogeneity of Immune
Infiltration--Determination of ITLR
[0213] An automated image analysis tool identified cancer,
lymphocytes and stromal cells encompassing fibroblasts and
endothelial cells based on their nuclear morphologies in H&E
whole-tumour section slides (Yuan et al. 2012). The main component
of this tool is a classifier trained by pathologists over randomly
selected tumour regions and validated in 564 breast tumours with
90% accuracy (Yuan et al. 2012). As a result of image analysis, the
types and spatial locations of on average 110,000 cells were
recorded in every breast tumour section. Thus, this fully automated
tool enabled the mapping of spatial distributions of all cancer
cells and lymphocytes within a tumour section, which can be
subsequently visualised as a 3D landscape (FIG. 1A). The spatial
relationships of immune and cancer cells are then analysed with a
statistical pipeline exemplified in FIG. 1B. First, to globally
profile the spatial distribution of the cancer cells, the cancer
cell density was quantified using a kernel estimate (Methods).
Intuitively, this builds a `cancer landscape` where hills indicate
tumour regions densely populated with cancer cells. The height of a
hill thus correlates with cancer density at a specific location in
the tumour (FIG. 1B). Secondly, for every lymphocyte, its spatial
proximity to cancer can be directly quantified with the cancer
density landscape at its specific location. Thus a quantitative
measurement of the spatial proximity to tumour cells can be
efficiently obtained for every lymphocyte (FIG. 1B).
[0214] Using this approach, we quantified the spatial proximity to
cancer for every lymphocyte in 181 TNBC samples in the METABRIC
study (Methods, FIG. 2A). In principle, lymphocytes that differ in
their spatial positioning to cancer can be differentiated based on
these quantitative spatial measurements. The inventor investigated
whether data-driven clustering methods based on normal distribution
can be used to differentiate different classes of lymphocytes,
since cell spatial distribution is a naturally emerged pattern.
Unsupervised Gaussian Mixture Model clustering Fraley, 2003) was
employed to identify lymphocyte clusters based on their spatial
proximity to cancer using a training set of 100,000 randomly
sampled lymphocytes (FIG. 2B, Methods). Subsequently, a
three-cluster solution that identify three classes of lymphocytes
was considered the optimal by the Bayesian Information Criterion
(Schwartz, 1978) (FIG. 2B). This three-class solution is
consistently the optimal 97% of the time upon 200 repeated
sampling, whilst the five-class solution was considered optimal 3%
of the time (Methods, FIG. 2C). In addition, the cluster structure
of the three-class solutions was stable (median of cluster mean:
0.011, 0.06, 0.13; standard deviation/SD: 0.002, 0.0047, 0.0045;
FIG. 2C), indicating that the same clusters were identified in each
random sampling. The three classes of lymphocytes were named as
Intra-Tumour Lymphocyte (ITL), Adjacent-Tumour Lymphocyte (ATL) and
Distal-Tumour Lymphocyte (DTL). Subsequently, a classifier was
trained based on the lymphocyte classes to predict the types of
lymphocytes in all TNBC samples (Methods).
[0215] To understand the differences of the newly proposed
lymphocyte classes, additional measures were derived that are based
on direct physical distances. First, for each lymphocyte its
distance to the nearest cancer cell can be quantified (d.sub.min,
Methods, FIG. 2D). It was shown that ITLs have a median distance of
7 .mu.m (interquartile range 5-10) to the nearest cancer cell,
whilst it is 10 .mu.m (7-11) for ATLs, and 20 .mu.m (14-26) for
DTLs (FIG. 2E). The overlap in distance to nearest cancer cell
between ITLs and ATLs suggests that this measure is not the
fundamental difference between the two classes. Since the kernel
density measure based on which the lymphocyte classes were derived
is essentially spatial smoothing, the inventor hypothesised that
the spatial arrangement of cancer cells surrounding lymphocytes
differs between ATLs and ITLs. To measure spatial arrangement, the
inventor examined the convex hull region formed by 5 nearest cancer
cells, which is the smallest region that covers these cells (FIG.
2D, Methods). If a lymphocyte is surrounded by cancer cells, it
should fall into the convex hull region formed by nearby cancer
cells and has a small distance to the centroid of this region (FIG.
2D, left). In contrast, if nearby cancer cells are to one side of a
lymphocyte, the distance between the lymphocyte and the centroid of
the cancer convex hull region is likely to be large (FIG. 2D,
right). Thus, the inventor used the distance between a lymphocyte
and the centroid of the cancer convex hull region as a quantitative
measure of the spatial arrangement of cancer cells surrounding a
lymphocyte (d.sub.centroid) Three lymphocyte classes displayed
significant differences in d.sub.centroid with median
d.sub.centroid 3.6 .mu.m (2.2-5.1), 7.2 .mu.m (4.5-10.6), 17.7
.mu.m (11.0-26.6) for ITLs, ATLs, and DTLs, respectively (FIG. 2E).
Therefore, d.sub.mind and d.sub.centroid together better define and
aid interpretation of the lymphocyte classes (FIG. 2F). Taken
together, these data demonstrated that the proposed kernel-based
measure of spatial proximity to cancer can effectively account for
spatial proximity and surroundings, and also that the three
lymphocyte classes differ not only in the distance to the nearest
cancer cell but also in the ways nearby cancer cells are arranged.
A representative case showing spatial distribution of lymphocytes
in these three classes is illustrated (FIG. 3A-B). For instance,
the ITLs can be observed to locate within regions densely populated
with cancer cells (FIG. 3C).
[0216] In the 181 TNBC samples, there are overall more ATLs than
the other two types of lymphocytes (on average 47% ATLs, 32% ITLs
and 21% DTLs, FIG. 4A). The changes in abundance of these three
classes in 181 samples can be observed in a triangle plot (FIG.
4B). When the proportion of ITLs is low (0-20%), there are in
general more DTLs (40-60%) than ATLs (30-50%). As the amount the
ITLs increase (20-50%), ATLs also increase (40-60%) while DTLs
decrease (10-40%). When there are large amount of ITLs (>50%),
there are still certain amount of ATLs (20-40%) with very few DTLs
(<10%). To summarise the degree of lymphocytic infiltration for
a given tumour, we first calculated the ratio between the number of
ITLs and the number of cancer cells (ITLR; see Methods above). In
the 181 TNBC samples, a significant association was observed
between ITLR and pathological assessment of lymphocytic
infiltration of the tumours in categories of absent, mild and
severe (p=2.times.10.sup.-33, FIG. 4C). In terms of other clinical
parameters, there was no correlation between ITLR and tumour size,
node status and TP53 mutation status (FIG. 4D). Tumour grade was
not considered because 87% of the TNBC samples are Grade 3 tumours.
These data support ITLR's validity as a measurement of lymphocytic
infiltration and its potential value in addition to known clinical
parameters for TNBC.
[0217] ITLR is a Statistical Measure of Lymphocytic Infiltration
and an Independent Predictor of Disease-Specific Survival in Two
TNBC Cohorts.
[0218] To investigate the clinical significance of the proposed
immune measure ITL, the inventor analysed disease-specific survival
as a function of ITL. The TNBC samples can be divided into two
independent cohorts based on contributing hospitals (Methods, n=89
and n=92, distribution of ITLR FIG. 3E). To dichotomise the
continuous ITLR, the optimal cut-off was selected to have the best
prognostic value in Cohort 1 as the discovery cohort (Methods). The
best cut-off was selected to be 0.011 and 20% of the patients have
ITLR lower than this cut-off. These patients have significantly
worse disease-specific survival compared with patients with higher
ITLR in Cohort 1 (Log-rank test p=0.0063, Hazard ratio HR=0.36, 95%
confidence interval CI=0.17-0.77; Table 1; FIG. 4F). This
observation was verified in the validation cohort, Cohort 2
(p=0.0037, HR=0.25, CI=0.09-0.69; FIG. 4F). Significant
stratification was observed upon repeated analysis with Cohort 2 as
the discovery and Cohort 1 as the validation cohort (FIG. 3G). The
same tests were performed for the ratio of ATLs and DTLs to cancer
cells (ATLR and DTLR), but neither showed a significant correlation
with disease-specific survival (Discovery and Validation cohort:
ATLR p=0.064 and 0.75; DTLR p=0.43 and 0.25; FIG. 7-8). We
subsequently focused on ITLR. ITLR-high TNBC patients have a
survival probability of 80% five year from diagnosis versus 49% for
ITLR-low patients (Kaplan-Meier survival estimates, two cohorts
combined).
[0219] ITLR was compared with eight other immune signatures. These
include the previously published image-based signature, lymphocyte
abundance (Lym), defined as the ratio between the number of
lymphocytes and the number of cancer cells (Methods) (Yuan et al.
2012). A major difference between ITLR and Lym is that Lym does not
account for different classes of lymphocytes whilst ITLR considers
infiltrating lymphocytes. The remainder of signatures are published
gene expression-based signatures from Calabro et al. (Calabro et
al.) that is predictive of ER-negative breast cancer prognosis, a
5-gene signature from Ascierto et al. (Ascierto et al. '12) that
predicts recurrence-free survival across breast cancer subtypes,
and the B-cell, IL8 and combined signatures to predict prognosis of
TNBC (Rody et al.). CXCR3 and CXCL13 expression were also included
since they have been shown to correlate with breast cancer
prognosis (Ma et al., Gu-Trantien et al.).
[0220] The same cut-off selection approach was applied to test the
association between these signatures and disease-specific survival
(Table 2). The signatures that showed the best prognostic values
are shown in FIG. 5A-E (all are provided in FIG. 9) and Table 1.
None of these signatures correlated with prognosis in both cohorts.
This analysis was repeated using Cohort 2 as the discovery cohort
for selecting the optimal cut-offs and Cohort 1 for validation
(FIG. 10, Table 3). In both experiments, only ITLR consistently
stratified patients into two groups of different outcome among the
nine signatures (FIGS. 7 and 8). Furthermore, the best ITLR
cut-offs selected in two cohorts for all nine signatures were
compared (Methods, FIG. 5F). ITLR was among the most consistent
signatures in terms of optimal cut-offs in two cohorts, supporting
the consistency and the potential use of ITLR as an objective
measure for identifying patients with low lymphocytic
infiltration.
[0221] Compared to published immune signatures, ITLR was also the
only signature to show significant correlation with
disease-specific survival in multivariate Cox proportional hazards
model together with standard clinical parameters of nodal status
and tumour size in both cohorts, whichever cohort was used as the
discovery cohort (Tables 1 to 3). Using samples from both cohorts,
ITLR has a log-rank p-value of 2.1.times.10.sup.-4 and HR 0.32
(0.17-0.58). To test the robustness of the Cox model in determining
the prognostic value of ITLR, we used bootstrap analysis in
randomly perturbed data and repeated the univariate and
multivariate regression analysis 1,000 times. In 95.6% and 94.7% of
the time, ITLR remained significantly associated with prognosis in
univariate and multivariate analysis, respectively. Taken together,
these results show the stability and robustness of ITLR as an
independent prognostic biomarker in TNBC.
[0222] ITLR Heterogeneity is Reflected on the Transcriptional Level
by CTLA4 and APOBEC3G Expression
[0223] To identify molecular associations of immune infiltration
and to test the biological relevance of ITLR, the inventor
integrated image-based ITLR with microarray gene expression data
profiled for the same set of 181 TNBC tumours. The analysis
identified 307 genes positively correlated and 105 genes negatively
correlated with ITLR (False Discovery Rate multiple testing
correction, q-value<0.05; Methods). Genes with the most
significant correlations with our immune signature ITLR include
kinases (SH3KBP1, LCK, MAP4K1) and receptors (FCRL3, GPR18,
TNFRSF13B, SEMA4D, CXCR3, IL2RG), as well as the known
immunotherapy target CTLA4 (Table 4). Thus, significant
correlations between ITLR and immune-related genes demonstrate the
biological relevance of the ITLR signature.
TABLE-US-00009 TABLE 4 Top 20 genes positively correlated with ITLR
and top 10 genes negatively correlated with ITLR (underline).
Symbol Cytoband Description cor q SH3KBP1 Xp22.12b SH3-domain
kinase binding protein 1 0.4 0.0011 FCRL3 1q23.1d Fc receptor-like
3 0.4 0.0011 LCK 1p35.1b lymphocyte-specific protein tyrosine
kinase 0.4 0.0011 GPR18 13q32.3a G protein-coupled receptor 18 0.39
0.0011 TNFRSF13B 17p11.2h tumour necrosis factor receptor
superfamily, 0.39 0.0011 member 13B SEMA4D/ 9q22.2a sema domain,
immunoglobulin domain (Ig), 0.39 0.0012 CD100 transmembrane domain
(TM) and short cytoplasmic domain, (semaphorin) 4D MAP4K1 19q13.2a
mitogen-activated protein kinase kinase kinase 0.39 0.0012 kinase 1
RLTPR 16q22.1b RGD motif, leucine rich repeats, tropomodulin 0.38
0.0012 domain and proline-rich containing UBASH3A 21q22.3b
ubiquitin associated and SH3 domain 0.38 0.0012 containing A IKZF3
17q12c IKAROS family zinc finger 3 (Aiolos) 0.38 0.0012 CYFIP2
5q33.3a- cytoplasmic FMR1 interacting protein 2 0.38 0.0012 q33.3b
CXCR3 Xq13.1d chemokine (C-X-C motif) receptor 3 0.38 0.0012 CD3E
11q23.3d CD3e molecule, epsilon (CD3-TCR complex) 0.38 0.0012 IL2RG
Xq13.1c interleukin 2 receptor, gamma 0.38 0.0012 CXCR5 11q23.3e
chemokine (C-X-C motif) receptor 5 0.38 0.0014 CTSW 11q13.1d
cathepsin W 0.37 0.0018 SH2D1A Xq25c SH2 domain containing 1A 0.37
0.0018 SEPT6 Xq24c septin 6 0.37 0.0018 CTLA4 2q33.2a cytotoxic
T-lymphocyte-associated protein 4 0.37 0.0019 SIRPG 20p13e
signal-regulatory protein gamma 0.37 0.0019 C10orf141 10q26.2b -0.4
0.0011 CD151 11p15.5c CD151 molecule (Raph blood group) -0.39
0.0011 SPP1 4q22.1b secreted phosohoprotein 1 -0.39 0.0012 ANXA2
15q22.2a annexin A2 -0.39 0.0012 P4HA2 5q31.1b prolyl
4-hydroxylase, alpha polypeptide II -0.36 0.0022 MUSK 9q31.3b
muscle, skeletal, receptor tyrosine kinase -0.36 0.0023 POFUT2
21q22.3e protein O-fucosyltransferase 2 -0.36 0.0025 ITGB5 3q21.2a
integrin, beta 5 -0.35 0.004 MXRA7 17q25.1d- matrix-remodelling
associated 7 -0.34 0.0046 q25.2a CALN1 7q11.22c calneuron 1 -0.34
0.0046
[0224] Subsequently, enrichment analysis was performed on the
positively and negatively correlated genes respectively against
MSigDB gene set categories (Subramanian, 2005) including KEGG
pathways (Kanehisa, 2000), canonical pathways curated by domain
experts and immunologic signatures (Methods, FIG. 11). Genes
positively correlated with ITLR are enriched with natural killer
cell mediated cytotoxicity, T cell receptor, Antigen processing and
presentation KEGG pathways, CD8 T cell, CD4 T cell and B cell
up-regulated immunogenic signatures, as well as IL12 and CD8 TCR
canonical pathways. Conversely, genes negatively correlated with
ITLR were enriched with ECM receptor interaction and focal adhesion
KEGG pathways, regulatory T cell and TGF.beta. related immunologic
signatures as well as integrin related pathways. The molecular
analysis on the pathway level suggests ITLR is positively
associated with anti-tumour immune activities in TNBC.
[0225] To further dissect their interconnected relationships and
discover de novo molecular modules, tightly connected gene modules
were identified within ITLR-associated genes (FIG. 11; Methods). As
such, seven modules of positively correlated genes (P1-7) and two
modules of genes negatively correlated with ITLR (N1 and N2) were
identified. Known immune-related genes in the modules include IFNG
(P1), RLPTR (P3), GPR18 (P4), CXCR3 (P5), MAP4K1 (P6), CTLA4 (P7),
ANXA2 (N1) and FAP (N2). Notably, two of the modules contain
APOBEC3G (P2) and CTLA4 (P7), which may suggest co-regulation among
APOBEC3G, NKG7 and interleukins including IL21R and IL18RAP, as
well as high correlations among CTLA4, chemoattractant for B
lymphocytes CXCL13 (Denkert) and TIGIT T cell immunoreceptor with
Ig and ITIM domains. Furthermore, expression profiles of these
genes were significantly associated with disease-specific survival
in TNBC, including APOBEC3G as well as GPR18 (P4) and MAP4K1 (P6)
ranked as the top ITLR-associated genes (FIG. 6B, FIG. 12). CTLA4
expression was able to stratify patients into groups with
significantly different prognosis, and could further stratify the
ITLR high group into two subgroups with significantly different
outcome (p=0.046, FIG. 6C, FIG. 12). Comparing ITLR with
ITLR-associated genes in terms of prognostic value, multivariate
analysis showed that ITLR stratification has additional and in many
cases superior value to ITLR-associated genes (FIG. 13,
Methods).
REFERENCES
[0226] A number of publications are cited above in order to more
fully describe and disclose the invention and the state of the art
to which the invention pertains. Full citations for these
references are provided below. The entirety of each of these
references is incorporated herein. [0227] Alexiou G A,
Vartholomatos E, Voulgaris S. Prognostic value of
neutrophil-to-lymphocyte ratio in patients with glioblastoma.
Journal of neuro-oncology. 2013; 115:521-2. [0228] Ancuta E, Ancuta
C, Zugun-Eloae F, Iordache C, Chirieac R, Carasevici E. Predictive
value of cellular immune response in cervical cancer. Romanian
journal of morphology and embryology=Revue roumaine de morphologie
et embryologie. 2009; 50:651-5. [0229] Andre, F., M. V. Dieci, P.
Dubsky, C. Sotiriou, G. Curigliano, C. Denkert and S. Loi (2013).
"Molecular pathways: involvement of immune pathways in the
therapeutic response and outcome in breast cancer." Clin Cancer Res
19(1): 28-33. [0230] Ascierto, M. L., M. Kmieciak, M. O. Idowu, R.
Manjili, Y. Zhao, M. Grimes, C. Dumur, E. Wang, V. Ramakrishnan, X.
Y. Wang, H. D. Bear, F. M. Marincola and M. H. Manjili (2012). "A
signature of immune function genes associated with recurrence-free
survival in breast cancer patients." Breast Cancer Res Treat
131(3): 871-880. [0231] Bambury R M, Teo M Y, Power D G, Yusuf A,
Murray S, Battley J E, et al. The association of pre-treatment
neutrophil to lymphocyte ratio with overall survival in patients
with glioblastoma multiforme. Journal of neuro-oncology. 2013;
114:149-54 [0232] Basavanhally, A. N., S. Ganesan, S. Agner, J. P.
Monaco, M. D. Feldman, J. E. Tomaszewski, G. Bhanot and A.
Madabhushi (2010). "Computerized image-based detection and grading
of lymphocytic infiltration in HER2+ breast cancer histopathology."
IEEE Trans Biomed Eng 57(3): 642-653. [0233] Berman, M. and P.
Diggle (1989). "Estimating Weighted Integrals of the 2nd-Order
Intensity of a Spatial Point Process." Journal of the Royal
Statistical Society Series B-Methodological 51(1): 81-92. [0234]
Jonckheere, A. R. (1954). "A test of significance for the relation
between m rankings and k ranked categories." British Journal of
Statistical Psychology 7: 93-100. [0235] Bystryn J C, Oratz R,
Roses D, Harris M, Henn M, Lew R. Relationship between immune
response to melanoma vaccine immunization and clinical outcome in
stage II malignant melanoma. Cancer. 1992; 69:1157-64. [0236]
Calabro, A., T. Beissbarth, R. Kuner, M. Stojanov, A. Benner, M.
Asslaber, F. Ploner, K. Zatloukal, H. Samonigg, A. Poustka and H.
Sultmann (2009). "Effects of infiltrating lymphocytes and estrogen
receptor on gene expression and prognosis in breast cancer." Breast
Cancer Res Treat 116(1): 69-77. [0237] Cheng, Yizong (August 1995).
"Mean Shift, Mode Seeking, and Clustering". IEEE Transactions on
Pattern Analysis and Machine Intelligence (IEEE) 17 (8): 790-799.
[0238] Clarke B, Tinker A V, Lee C H, Subramanian S, van de Rijn M,
Turbin D, et al. Intraepithelial T cells and prognosis in ovarian
carcinoma: novel associations with stage, tumor type, and BRCA1
loss. Modern pathology: an official journal of the United States
and Canadian Academy of Pathology, Inc. 2009; 22:393-402. [0239]
Contardi, E., G. L. Palmisano, P. L. Tazzari, A. M. Martelli, F.
Fala, M. Fabbi, T. Kato, E. Lucarelli, D. Donati, L. Polito, A.
Bolognesi, F. Ricci, S. Salvi, V. Gargaglione, S. Mantero, M.
Alberghini, G. B. Ferrara and M. P. Pistillo (2005). "CTLA-4 is
constitutively expressed on tumor cells and can trigger apoptosis
upon ligand interaction." Int J Cancer 117(4): 538-550. [0240]
Curtis, C., S. P. Shah, S. F. Chin, G. Turashvili, O. M. Rueda, M.
J. Dunning, D. Speed, A. G. Lynch, S. Samarajiwa, Y. Yuan, S. Graf,
G. Ha, G. Haffari, A. Bashashati, R. Russell, S. McKinney, A.
Langerod, A. Green, E. Provenzano, G. Wishart, S. Pinder, P.
Watson, F. Markowetz, L. Murphy, I. Ellis, A. Purushotham, A. L.
Borresen-Dale, J. D. Brenton, S. Tavare, C. Caldas and S. Aparicio
(2012). "The genomic and transcriptomic architecture of 2,000
breast tumours reveals novel subgroups." Nature 486(7403): 346-352.
[0241] Crane C A, Austgen K, Haberthur K, Hofmann C, Moyes K W,
Avanesyan L, et al. Immune evasion mediated by tumor-derived
lactate dehydrogenase induction of NKG2D ligands on myeloid cells
in glioblastoma patients. Proceedings of the National Academy of
Sciences of the United States of America. 2014; 111:12823-8. [0242]
Dave S S, Wright G, Tan B, Rosenwald A, Gascoyne R D, Chan W C, et
al. Prediction of Survival in Follicular Lymphoma Based on
Molecular Features of Tumor-Infiltrating Immune Cells. New England
Journal of Medicine. 2004; 351:2159-69. [0243] Davidsson S, Ohlson
A L, Andersson S O, Fall K, Meisner A, Fiorentino M, et al. CD4
helper T cells, CD8 cytotoxic T cells, and FOXP3(+) regulatory T
cells with respect to lethal prostate cancer. Modern pathology: an
official journal of the United States and Canadian Academy of
Pathology, Inc. 2013; 26:448-55. [0244] DeNardo, D. G., D. J.
Brennan, E. Rexhepaj, B. Ruffell, S. L. Shiao, S. F. Madden, W. M.
Gallagher, N. Wadhwani, S. D. Keil, S. A. Junaid, H. S. Rugo, E. S.
Hwang, K. Jirstrom, B. L. West and L. M. Coussens (2011).
"Leukocyte complexity predicts breast cancer survival and
functionally regulates response to chemotherapy." Cancer Discov
1(1): 54-67. [0245] Denkert, C., S. Loibl, A. Noske, M. Roller, B.
M. Muller, M. Komor, J. Budczies, S. Darb-Esfahani, R. Kronenwett,
C. Hanusch, C. von Torne, W. Weichert, K. Engels, C. Solbach, I.
Schrader, M. Dietel and G. von Minckwitz (2010). "Tumor-associated
lymphocytes as an independent predictor of response to neoadjuvant
chemotherapy in breast cancer." J Clin Oncol 28(1): 105-113. [0246]
Dieci, M. V., C. Criscitiello, A. Goubar, G. Viale, P. Conte, V.
Guarneri, G. Ficarra, M. C. Mathieu, S. Delaloge, G. Curigliano and
F. Andre (2014). "Prognostic value of tumor-infiltrating
lymphocytes on residual disease after primary chemotherapy for
triple-negative breast cancer: a retrospective multicenter study."
Ann Oncol 25(3): 611-618. [0247] Dieu-Nosjean M C, Antoine M
Fau-Danel C, Danel C Fau-Heudes D, Heudes D Fau-Wislez M, Wislez M
Fau-Poulot V, Poulot V Fau-Rabbe N, et al. Long-term survival for
patients with non-small-cell lung cancer with intratumoral lymphoid
structures. [0248] Erdag G, Schaefer J T, Smolkin M E, Deacon D H,
Shea S M, Dengel L T, et al. Immunotype and immunohistologic
characteristics of tumor-infiltrating immune cells are associated
with clinical outcome in metastatic melanoma. Cancer research.
2012; 72:1070-80 [0249] Failmezger H, Yuan Y, Rueda O and Markowetz
F (2012). CRImage: CRImage a package to classify cells and
calculate tumour cellularity. R package version 1.14.0. [0250]
Fiorelli V, Gendelman R, Caterina Sirianni M, Chang H-K, Colombini
S, Markham P D, et al. .gamma.-Interferon Produced by CD8+ T Cells
Infiltrating Kaposi's Sarcoma Induces Spindle Cells With Angiogenic
Phenotype and Synergy With Human Immunodeficiency Virus-1 Tat
Protein: An Immune Response to Human Herpesvirus-8 Infection?
Blood. 1998; 91:956-67. [0251] Fraley, C. and A. E. Raftery (2003).
"Enhanced model-based clustering, density estimation, and
discriminant analysis software: MCLUST." Journal of Classification
20(2): 263-286. [0252] Fridman, W. H., F. Pages, C. Sautes-Fridman
and J. Galon (2012). "The immune contexture in human tumours:
impact on clinical outcome." Nat Rev Cancer 12(4): 298-306. [0253]
Gannon P O, Poisson A O, Delvoye N, Lapointe R, Mes-Masson A M,
Saad F. Characterization of the intra-prostatic immune cell
infiltration in androgen-deprived prostate cancer patients. Journal
of immunological methods. 2009; 348:9-17. [0254] Galon, J., A.
Costes, F. Sanchez-Cabo, A. Kirilovsky, B. Mlecnik, C.
Lagorce-Pages, M. Tosolini, M. Camus, A. Berger, P. Wind, F.
Zinzindohoue, P. Bruneval, P. H. Cugnenc, Z. Trajanoski, W. H.
Fridman and F. Pages (2006). "Type, density, and location of immune
cells within human colorectal tumors predict clinical outcome."
Science 313(5795): 1960-1964. [0255] Galon, J., B. Mlecnik, G.
Bindea, H. K. Angell, A. Berger, C. Lagorce, A. Lugli, I. Zlobec,
A. Hartmann, C. Bifulco, I. D. Nagtegaal, R. Palmqvist, G. V.
Masucci, G. Botti, F. Tatangelo, P. Delrio, M. Maio, L. Laghi, F.
Grizzi, M. Asslaber, C. D'Arrigo, F. Vidal-Vanaclocha, E. Zavadova,
L. Chouchane, P. S. Ohashi, S. Hafezi-Bakhtiari, B. G. Wouters, M.
Roehrl, L. Nguyen, Y. Kawakami, S. Hazama, K. Okuno, S. Ogino, P.
Gibbs, P. Waring, N. Sato, T. Torigoe, K. Itoh, P. S. Patel, S. N.
Shukla, Y. Wang, S. Kopetz, F. A. Sinicrope, V. Scripcariu, P. A.
Ascierto, F. M. Marincola, B. A. Fox and F. Pages (2014). "Towards
the introduction of the `Immunoscore` in the classification of
malignant tumours." J Pathol 232(2): 199-209. [0256] Gao Q, Qiu Sj
Fau-Fan J, Fan J Fau-Zhou J, Zhou J Fau-Wang X-Y, Wang Xy Fau-Xiao
Y-S, Xiao Ys Fau-Xu Y, et al. Intratumoral balance of regulatory
and cytotoxic T cells is associated with prognosis of
hepatocellular carcinoma after resection. [0257] Gu-Trantien, C.,
S. Loi, S. Garaud, C. Equeter, M. Libin, A. de Wind, M. Ravoet, H.
Le Buanec, C. Sibille, G. Manfouo-Foutsop, I. Veys, B. Haibe-Kains,
S. K. Singhal, S. Michiels, F. Rothe, R. Salgado, H. Duvillier, M.
Ignatiadis, C. Desmedt, D. Bron, D. Larsimont, M. Piccart, C.
Sotiriou and K. Willard-Gallo (2013). "CD4(+) follicular helper T
cell infiltration predicts breast cancer survival." J Clin Invest
123(7): 2873-2892. [0258] Halliday G, Patel A, Hunt M, Tefany F,
Barnetson R C. Spontaneous regression of human melanoma/nonmelanoma
skin cancer: Association with infiltrating CD4+ T cells. World J
Surg. 1995; 19:352-8. [0259] Hastie R, Hastie, R. Tibshirani, J.
Friedman, The Elements of Statistical Learning (Springer, New York,
2001). [0260] Hegmans J P, Hemmes A, Hammad H, Boon L, Hoogsteden H
C, Lambrecht B N. Mesothelioma environment comprises cytokines and
T-regulatory cells that suppress immune responses. The European
respiratory journal. 2006; 27:1086-95. [0261] Hiraoka N, Onozato K
Fau-Kosuge T, Kosuge T Fau-Hirohashi S, Hirohashi S. Prevalence of
FOXP3+ regulatory T cells increases during the progression of
pancreatic ductal adenocarcinoma and its premalignant lesions.
[0262] Huang C T, Oyang Y J, Huang H C, Juan H F. MicroRNA-mediated
networks underlie immune response regulation in papillary thyroid
carcinoma. [0263] Issa-Nummer, Y., S. Darb-Esfahani, S. Loibl, G.
Kunz, V. Nekljudova, I. Schrader, B. V. Sinn, H. U. Ulmer, R.
Kronenwett, M. Just, T. Kuhn, K. Diebold, M. Untch, F. Holms, J. U.
Blohmer, J. O. Habeck, M. Dietel, F. Overkamp, P. Krabisch, G. von
Minckwitz and C. Denkert (2013). "Prospective validation of
immunological infiltrate for prediction of response to neoadjuvant
chemotherapy in HER2-negative breast cancer--a substudy of the
neoadjuvant GeparQuinto trial." PLoS One 8(12): e79775. [0264]
Janowczyk A, S. C, Madabhushi A. Quantifying local heterogeneity
via morphologic scale: Distinguishing tumoral from stromal regions.
Journal of pathology informatics. 2013; 4(Suppl). [0265] Kanehisa,
M. and S. Goto (2000). "KEGG: kyoto encyclopedia of genes and
genomes." Nucleic Acids Research 28(1): 27-30. [0266] Kmiecik J,
Poli A, Brons N H, Waha A, Eide G E, Enger P O, et al. Elevated
CD3+ and CD8+ tumor-infiltrating immune cells correlate with
prolonged survival in glioblastoma patients despite integrated
immunosuppressive mechanisms in the tumor microenvironment and at
the systemic level. Journal of neuroimmunology. 2013; 264:71-83.
[0267] Kruger, J. M., C. Wemmert, L. Sternberger, C. Bonnas, G.
Dietmann, P. Gancarski and F. Feuerhake (2013). "Combat or
surveillance? Evaluation of the heterogeneous inflammatory breast
cancer microenvironment." J Pathol 229(4): 569-578. [0268] Kuong,
K. J. and L. A. Loeb (2013). "APOBEC3B mutagenesis in cancer." Nat
Genet 45(9): 964-965. [0269] Monajemi, M., C. F. Woodworth, J.
Benkaroun, M. Grant and M. Larijani (2012). "Emerging complexities
of APOBEC3G action on immunity and viral fitness during HIV
infection and treatment." Retrovirology 9: 35. [0270] Kono K,
Kawaida H, Takahashi A, Sugai H, Mimura K, Miyagawa N, et al.
CD4(+)CD25high regulatory T cells increase with tumor stage in
patients with gastric and esophageal cancers. Cancer Immunol
Immunother. 2006; 55:1064-71. [0271] Lee, H. J., J. Y. Seo, J. H.
Ahn, S. H. Ahn and G. Gong (2013). "Tumor-associated lymphocytes
predict response to neoadjuvant chemotherapy in breast cancer
patients." J Breast Cancer 16(1): 32-39. [0272] Liu, S., J.
Lachapelle, S. Leung, D. Gao, W. D. Foulkes and T. O. Nielsen
(2012). "CD8+ lymphocyte infiltration is an independent favorable
prognostic indicator in basal-like breast cancer." Breast Cancer
Res 14(2): R48. [0273] Loi, S., N. Sirtaine, F. Piette, R. Salgado,
G. Viale, F. Van Eenoo, G. Rouas, P. Francis, J. P. Crown, E.
Hitre, E. de Azambuja, E. Quinaux, A. Di Leo, S. Michiels, M. J.
Piccart and C. Sotiriou (2013). "Prognostic and predictive value of
tumor-infiltrating lymphocytes in a phase III randomized adjuvant
breast cancer trial in node-positive breast cancer comparing the
addition of docetaxel to doxorubicin with doxorubicin-based
chemotherapy: BIG 02-98." J Clin Oncol 31(7): 860-867. [0274]
McNamara M G, Lwin Z, Jiang H, Templeton A J, Zadeh G, Bernstein M,
et al. Factors impacting survival following second surgery in
patients with glioblastoma in the temozolomide treatment era,
incorporating neutrophil/lymphocyte ratio and time to first
progression. Journal of neuro-oncology. 2014; 117:147-52. [0275]
Ma, X., K. Norsworthy, N. Kundu, W. H. Rodgers, P. A. Gimotty, O.
Goloubeva, M. Lipsky, Y. Li, D. Holt and A. Fulton (2009). "CXCR3
expression is associated with poor survival in breast cancer and
promotes metastasis in a murine model." Mol Cancer Ther 8(3):
490-498. [0276] Mahmoud, S. M. A., E. C. Paish, D. G. Powe, R. D.
Macmillan, M. J. Grainge, A. H. S. Lee, I. O. Ellis and A. R. Green
(2011). "Tumor-infiltrating CD8+ lymphocytes predict clinical
outcome in breast cancer." J Clin Oncol 29(15): 1949-1955. [0277]
Mangeat, B., P. Turelli, G. Caron, M. Friedli, L. Perrin and D.
Trono (2003). "Broad antiretroviral defence by human APOBEC3G
through lethal editing of nascent reverse transcripts." Nature
424(6944): 99-103. [0278] Mao, H., L. Zhang, Y. Yang, W. Zuo, Y.
Bi, W. Gao, B. Deng, J. Sun, Q. Shao and X. Qu (2010). "New
insights of CTLA-4 into its biological function in breast cancer."
Curr Cancer Drug Targets 10(7): 728-736. [0279] Milne K, Kobel M,
Kalloger S E, Barnes R O, Gao D, Gilks C B, et al. Systematic
analysis of immune infiltrates in high-grade serous ovarian cancer
reveals CD20, FoxP3 and TIA-1 as positive prognostic factors. PloS
one. 2009; 4:e6412. [0280] Mittendorf et al (2014) "PD-L1
Expression in Triple Negative Breast Cancer" Cancer Immunol Res 2;
361. [0281] Mukherji B Fau-Guha A, Guha A Fau-Loomis R, Loomis R
Fau-Ergin M T, Ergin M T. Cell-mediated amplification and down
regulation of cytotoxic immune response against autologous human
cancer.
[0282] Natrajan, R., F. M. Mardakheh, H. Sailem, C. Bakal and Y.
Yuan (In preparation). "The Ecosystem Diversity Landscape of Breast
Cancer." [0283] Ohno S, Ohno Y, Suzuki N, Kamei T, Koike K, Inagawa
H, et al. Correlation of histological localization of
tumor-associated macrophages with clinicopathological features in
endometrial cancer. Anticancer research. 2004; 24:3335-42 [0284]
Papewalis C, Kouatchoua C Fau-Ehlers M, Ehlers M Fau-Jacobs B,
Jacobs B Fau-Porwol D, Porwol D Fau-Schinner S, Schinner S
Fau-Willenberg H S, et al. Chromogranin A as potential target for
immunotherapy of malignant pheochromocytoma.
[0285] Perou C, Sorlie T, Eisen M, van de Rijn M, Jeffrey S, Rees
C, et al. Molecular portraits of human breast tumours. Nature.
2000; 406:747-52. [0286] Prat, J., "Staging classification for
cancer of the ovary, fallopian tube and peritoneum." International
Journal of Gynecology and Obstetrics. 2013.
(http://www.ncin.org.uk/view?rid=2515). [0287] Pretscher D, Distel
L V, Grabenbauer G G, Wittlinger M, Buettner M, Niedobitek G.
Distribution of immune cells in head and neck cancer: CD8+ T-cells
and CD20+B-cells in metastatic lymph nodes are associated with
favourable outcome in patients with oro- and hypopharyngeal
carcinoma. BMC cancer. 2009; 9:292. [0288] Robert, C., L. Thomas,
I. Bondarenko, S. O'Day, D. J. M, C. Garbe, C. Lebbe, J. F.
Baurain, A. Testori, J. J. Grob, N. Davidson, J. Richards, M. Maio,
A. Hauschild, W. H. Miller, Jr., P. Gascon, M. Lotem, K.
Harmankaya, R. Ibrahim, S. Francis, T. T. Chen, R. Humphrey, A.
Hoos and J. D. Wolchok (2011). "Ipilimumab plus dacarbazine for
previously untreated metastatic melanoma." N Engl J Med 364(26):
2517-2526. [0289] Rody, A., U. Holtrich, L. Pusztai, C. Liedtke, R.
Gaetje, E. Ruckhaeberle, C. Solbach, L. Hanker, A. Ahr, D. Metzler,
K. Engels, T. Karn and M. Kaufmann (2009). "T-cell metagene
predicts a favorable prognosis in estrogen receptor-negative and
HER2-positive breast cancers." Breast Cancer Res 11(2): R15. [0290]
Rody A, Karn T, Liedtke C, Pusztai L, Ruckhaeberle E, Hanker L, et
al. A clinically relevant gene signature in triple negative and
basal-like breast cancer. Breast cancer research: BCR. 2011; 13:R97
[0291] Salgardo, R. et al., (2014) "Harmonization of the evaluation
of tumour infiltration lymphocytes (TILs) in breast cancer:
recommendations by an international TILs-Working Group 2014. Annals
of Oncology [0292] Salvi, S., V. Fontana, S. Boccardo, D. F. Merlo,
E. Margallo, S. Laurent, A. Morabito, E. Rijavec, M. G. Dal Bello,
M. Mora, G. B. Ratto, F. Grossi, M. Truini and M. P. Pistillo
(2012). "Evaluation of CTLA-4 expression and relevance as a novel
prognostic factor in patients with non-small cell lung cancer."
Cancer Immunol Immunother 61(9): 1463-1472. [0293] Schalper, K. A.,
V. Velcheti, D. Carvajal, H. Wimberly, J. Brown, L. Pusztai and D.
L. Rimm (2014). "In Situ Tumor PD-L1 mRNA Expression Is Associated
with Increased TILs and Better Outcome in Breast Carcinomas." Clin
Cancer Res. [0294] Spanos W C, Nowicki P, Lee D W, Hoover A,
Hostager B, Gupta A, et al. Immune response during therapy with
cisplatin or radiation for human papillomavirus-related head and
neck cancer. Archives of otolaryngology--head & neck surgery.
2009; 135:1137-46. [0295] Sjodahl G, Lovgren K, Lauss M, Chebil G,
Patschan O, Gudjonsson S, et al. Infiltration of CD3+ and CD68+
cells in bladder cancer is subtype specific and affects the outcome
of patients with muscle-invasive tumors. Urologic Oncology:
Seminars and Original Investigations. 2014; 32:791-7. [0296] Sorbye
S W, Kilvaer T, Valkov A, Donnem T, Smeland E, Al-Shibli K, et al.
Prognostic impact of lymphocytes in soft tissue sarcomas. PloS one.
2011; 6:e14611. [0297] Stagg, J. and B. Allard (2013).
"Immunotherapeutic approaches in triple-negative breast cancer:
latest research and clinical prospects." Ther Adv Med Oncol 5(3):
169-181. [0298] Subramanian, A., P. Tamayo, V. K. Mootha, S.
Mukherjee, B. L. Ebert, M. A. Gillette, A. Paulovich, S. L.
Pomeroy, T. R. Golub, E. S. Lander and J. P. Mesirov (2005). "Gene
set enrichment analysis: a knowledge-based approach for
interpreting genome-wide expression profiles." Proc Natl Acad Sci
USA 102(43): 15545-15550. [0299] Suzuki K, Kadota K, Sima C S,
Nitadori J, Rusch V W, Travis W D, et al. Clinical impact of immune
microenvironment in stage I lung adenocarcinoma: tumor
interleukin-12 receptor beta2 (IL-12Rbeta2), IL-7R, and stromal
FoxP3/CD3 ratio are independent predictors of recurrence. Journal
of clinical oncology: official journal of the American Society of
Clinical Oncology. 2013; 31:490-8. [0300] The Cancer Genome Atlas
Network (2012). "Comprehensive molecular portraits of human breast
tumours." Nature 490(7418): 61-70. [0301] Tjin E P, Krebbers G,
Meijlink K J, van de Kasteele W, Rosenberg E H, Sanders J, et al.
Immune-escape markers in relation to clinical outcome of advanced
melanoma patients following immunotherapy. Cancer immunology
research. 2014; 2:538-46. [0302] Thompson R H, Dong H, Lohse C M,
Leibovich B C, Blute M L, Cheville J C, et al. PD-1 is expressed by
tumor-infiltrating immune cells and is associated with poor outcome
for patients with renal cell carcinoma. Clinical cancer research:
an official journal of the American Association for Cancer
Research. 2007; 13:1757-61. [0303] Townsend K N, Spowart J E,
Huwait H, Eshragh S, West N R, Elrick M A, et al. Markers of T cell
infiltration and function associate with favorable outcome in
vascularized high-grade serous ovarian carcinoma. PloS one. 2013;
8:e82406. [0304] Ueno T, Toi M, Saji H, Muta M, Bando H, Kuroi K,
et al. Significance of macrophage chemoattractant protein-1 in
macrophage recruitment, angiogenesis, and survival in human breast
cancer. Clinical cancer research: an official journal of the
American Association for Cancer Research. 2000; 6:3282-9. [0305]
Vauleon E, Tony A, Hamlat A, Etcheverry A, Chiforeanu D C, Menei P,
et al. Immune genes are associated with human glioblastoma
pathology and patient survival. BMC medical genomics. 2012; 5:41.
[0306] Villegas F R, Coca S, Villarrubia V G, Jimenez R, Chilion M
J, Jareno J, et al. Prognostic significance of tumor infiltrating
natural killer cells subset CD57 in patients with squamous cell
lung cancer. Lung cancer (Amsterdam, Netherlands). 2002; 35:23-8.
[0307] Webster W S, Lohse C M, Thompson R H, Dong H, Frigola X,
Dicks D L, et al. Mononuclear cell infiltration in clear-cell renal
cell carcinoma independently predicts patient survival. Cancer.
2006; 107:46-53. [0308] Welsh T J, Green R H, Richardson D, Waller
D A, O'Byrne K J, Bradding P. Macrophage and mast-cell invasion of
tumor cell islets confers a marked survival advantage in
non-small-cell lung cancer. Journal of clinical oncology: official
journal of the American Society of Clinical Oncology. 2005;
23:8959-67. [0309] Witkin, A. P. "Scale-space filtering", Proc. 8th
Int. Joint Conf. Art. Intell., Karlsruhe, Germany, 1019-1022, 1983.
[0310] Wu G, Han B I Fau-Pei X-t, Pei X T. Immune response of
dendritic cells acquiring antigens from apoptotic
cholangiocarcinoma cells induced by mitomycin. [0311] Yang I, Tihan
T, Han S J, Wrensch M R, Wiencke J, Sughrue M E, et al. CD8+ T-cell
infiltrate in newly diagnosed glioblastoma is associated with
long-term survival. Journal of clinical neuroscience: official
journal of the Neurosurgical Society of Australasia. 2010;
17:1381-5. [0312] Yong A S, Stephens N, Weber G, Li Y, Savani B N,
Eniafe R, et al. Improved outcome following allogeneic stem cell
transplantation in chronic myeloid leukemia is associated with
higher expression of BMI-1 and immune responses to BMI-1 protein.
Leukemia. 2011; 25:629-37. [0313] Yuan, Y., H. Failmezger, 0. M.
Rueda, H. R. Ali, S. Graf, S.-F. Chin, R. F. Schwarz, C. Curtis, M.
J. Dunning, H. Bardwell, N. Johnson, S. Doyle, G. Turashvili, E.
Provenzano, S. Aparicio, C. Caldas and F. Markowetz (2012).
"Quantitative Image Analysis of Cellular Heterogeneity in Breast
Tumors Complements Genomic Profiling." Science Translational
Medicine 4(157): 157ra143. [0314] Zhang Y, Wang L, Liu Y, Wang S,
Shang P, Gao Y, et al. Preoperative neutrophil-lymphocyte ratio
before platelet-lymphocyte ratio predicts clinical outcome in
patients with cervical cancer treated with initial radical surgery.
International journal of gynecological cancer: official journal of
the International Gynecological Cancer Society. 2014;
24:1319-25.
* * * * *
References